Model Details
Contains details like Model name, a Project name to which a model belongs to, Category, Tags and brief description about the model.
Source Dataset
Input to a model can be flat file (.csv) or a dataset that belongs to the project.
Prepare Dataset -> Preview Data
Preview the dataset before proceeding for preprocessing and training a model.
Prepare Dataset -> Prediction Method -> Supervised
The supervised method is when the dataset has a target column to predict and train the model accordingly. EX: Classification, regression and forecasting method use a dependent column and predict that dependent (target) columns based on the patterns in all other columns.
Prepare Dataset -> Prediction Method -> UnSupervised
The unsupervised method is when the dataset doesn’t have a target column to predict model accordingly. EX: Clustering method used to cluster the data into multiple groups without any dependent column.
Prepare Dataset -> Prediction Method -> Problem type -> Classification
Classification method is used to predict a dependent (target) column based on all other columns. Where the dependent column has classes. Therefore, the target column is a categorical column.
Prepare Dataset -> Prediction Method -> Problem type -> Regression
Regression method is used to predict a dependent (target) column based on all other columns. Where the dependent column is a continuous numerical data. Therefore, the target column is a Numerical column.
Prepare Dataset -> Prediction Method -> Problem type -> Forecasting
The forecasting method predicts a dependent (target) column based on all other column date column and its frequency. Where the dependent column is a continuous numerical data. Therefore, the target column is a Numerical column.
Prepare Dataset -> Select Target column to Predict (Classification or regression)
User selects the dependent (target) column to predict based on objective of the model.
Prepare Dataset -> Select Time column to Predict (Forecasting)
User selects the Time column on which frequency the model is trained.
Prepare Dataset -> Select Freq and Group by Date column (Forecasting)
Group the dataset using various aggregate functions like ["mean", "median", "std", "sum", "min", "max", and "count"] on numerical features and using the frequency of the time column.
Prepare Dataset -> Select Target column to forecast (Forecasting)
User selects the dependent (target) column to forecast based on objective of the model.
Train-test Split Dataset -> Select Sampling Mode -> Random Sampling
Random sampling is a sampling technique where every sample has the same probability of being chosen.
Train-test Split Dataset -> Sampling split ratio
Ratio to split the data into train and test. Typically, Train dataset is more than test dataset.
Train-test Split Dataset-> Random Seed
The random seed is a value that ensures the use of the same sample for a particular input.
Train-test Split Dataset -> Validation Dataset
The validation dataset is used to validate the best model while tuning with hyperparameters.
Tarin-test Split Dataset -> Train Dataset
The Train dataset is used to validate the model using multiple algorithms.
Train-test Split Dataset -> Test Dataset
The Train dataset is used to validate the model using multiple algorithms.
Configure Train Dataset -> Select Features
Select features based on the feature importance and visualizations provided on the analyze data graphs.
Configure Train Dataset -> Treat Outliers
Treat Outliers is to remove or replace outliers in a few features of the dataset.
Configure Train Dataset -> Transform Train Dataset
Transforming dataset is to encode the categorical columns and scale the numerical columns.
Configure Train Dataset -> Transform Train Dataset -> Encoding
Encoding is to create a numerical value respectively for each category in the feature. EX : One-Hot Encoding is used for Nominal categories (categories which do not follow any hierarchy, for ex: arrival and destination city), Ordinal encoding is used for Ordinal data (categories which follow hierarchy for ex: ratings like... good, bad, neutral).
Configure Train Dataset -> Class Distribution
In the classification method of supervised learning, the dependent (target column) variable has multiple categories. These categories (classes) in the train dataset should have equal distribution so that the model is not biasly trained.
Configure Train Dataset -> Class Distribution -> Balance Data
The balance data button is used to create data using SMOTE which balances the number of samples of classes.
Configure Train Dataset -> Feature Selection
Feature selection screen is used to select only a few features of the dataset by dropping unnessasry features using various algorithms.
Configure Train Dataset -> Feature Selection -> Univariate
This works by selecting the best features based on univariate statistical tests like ANOVA F value , chi square and mutual information.
Configure Train Dataset -> Feature Selection -> Variance
This works by selecting the best features based on variance in the dataset.
Configure Train Dataset -> Feature Selection -> Multivariate
Univariate looks at each feature at a time and determines if it’s important. Multivariate looks at interactions as well. SelectFromModel and RFE (Recursive feature elimination) two methods used for multifvariate feature selection.
Configure Train Dataset -> Feature Selection -> Algorithm Type -> ANOVA F-Value Classification
Analysis of variance (ANOVA) can determine whether the means of three or more groups are different.
Configure Train Dataset -> Feature Selection -> Algorithm Type -> Mutual Information Classification
Measures the dependency between the variables dependent and indepenndent variables.
Configure Train Dataset -> Dimensionality Reduction
Transformation of data from a high-dimensional space into a low-dimensional space so that the low-dimensional representation retains some meaningful properties of the original data.
Configure Train Dataset -> Dimensionality Reduction -> PCA
Principal Component Analysis : Transforms a set of features in a dataset into a smaller number of features called principal components. Retains as much information in the original dataset as possible.
Configure Train Dataset -> Dimensionality Reduction -> SVD
Singular Value Decomposition: The idea behind the SVD is that a rectangular matrix can be broken down into a product of three other matrices that are easy to work with. This transformer performs linear dimensionality reduction by means of truncated singular value decomposition (SVD).
Configure Train Dataset -> Dimensionality Reduction -> Feature Agglomeration
Agglomerative clustering to group together features that look very similar, thus decreasing the number of features.
Configure Train Dataset -> Dimensionality Reduction -> Gaussian Random Projection
Reduces the dimensionality by projecting the original input space on a randomly generated matrix.
Configure Train Dataset -> Dimensionality Reduction -> Sparse Random Projection
Sparse Random Projection: reduces the dimensionality by projecting the original input space using a sparse random matrix. Sparse random matrices are an alternative to dense Gaussian random projection matrix that guarantees similar embedding quality while being much more memory efficient and allowing faster computation of the projected data.
Configure Train Dataset -> Dimensionality Reduction -> Number Of Components
Number of components to reduce into.
Evaluate Estimators -> Select Estimator
Estimators are multiple algorithms to use to train the machine learning model.
Evaluate Estimators -> Select Estimator -> Select Metrics
Metrics of the model to calculate.
Evaluate Estimators -> Select Estimator -> Select Metrics -> Classification-> Accuracy
Accuracy is the fraction of predictions our model got right. Formally, accuracy has the following definition: Accuracy = Number of correct Predictions / Total Number of predictions. For binary classification, accuracy can also be calculated in terms of positives and negatives as follows: Accuracy = (TP+TN)/(TP+TN+FP+FN). Where TP = True Positives, TN = True Negatives, FP = False Positives, and FN = False Negatives.
Evaluate Estimators -> Select Estimator -> Select Metrics -> Classification-> Precision
The proportion of positive identifications that were correct. Precision = TP / (TP+FP). Where TP = True Positives and FP = False Positives.
Evaluate Estimators -> Select Estimator -> Select Metrics -> Classification-> Recall
The proportion of actual positives that were identified correctly. Recall = TP / (TP+FN). Where TP = True Positives and FN = False Negatives.
Evaluate Estimators -> Select Estimator -> Select Metrics -> Classification-> F1 Score
The harmonic mean of precision and recall. In the F1 score, we compute the average of precision and recall. F1 Score = 2 * ((Precision * Recall)/ (Precision + Recall)). the F1 score gives equal weight to Precision and Recall.
Evaluate Estimators -> Select Estimator -> Select Metrics -> Classification-> Roc
A receiver operating characteristic curve, or ROC curve, is a graphical plot that illustrates the diagnostic ability of a binary classifier system as its discrimination threshold is varied.
Evaluate Estimators -> Select Estimator -> Select Metrics -> Regression-> R2 score
The R2 score function computes the coefficient of determination, usually denoted as R2. It represents the proportion of variance (of y) that has been explained by the independent variables in the model. It provides an indication of goodness of fit and therefore a measure of how well unseen samples are likely to be predicted by the model, through the proportion of explained variance.
Evaluate Estimators -> Select Estimator -> Select Metrics -> Regression-> max error
The max error function computes the maximum residual error, a metric that captures the worst-case error between the predicted value and the true value. In a perfectly fitted single output regression model, max_error would be 0 on the training set and though this would be highly unlikely in the real world, this metric shows the extent of error that the model had when it was fitted.
Evaluate Estimators -> Select Estimator -> Select Metrics -> Regression-> Mean Absolute error
The magnitude of difference between the prediction of observation and the true value of that observation.
Evaluate Estimators -> Select Estimator -> Select Metrics -> Regression-> Mean Squared error
The average of the square of the difference between the original values and the predicted values. The Mean Squared Error measures how close a regression line is to a set of data points. The lower the value the better and 0 means the model is perfect. R-Squared is also termed the standardized version of MSE.
Evaluate Estimators -> Select Estimator -> Select Metrics -> Regression-> Median Absolute error
The median absolute error is particularly interesting because it is robust to outliers. The loss is calculated by taking the median of all absolute differences between the target and the prediction.
Evaluate Estimators -> Select Estimator -> Select Metrics -> Regression-> Mean squared log error
The mean squared log error function computes a risk metric corresponding to the expected value of the squared logarithmic (quadratic) error or loss.
Evaluate Estimators -> Select Estimator -> Select Metrics -> Regression-> Mean absolute percentage error
The mean absolute percentage error (MAPE), also known as mean absolute percentage deviation (MAPD), is an evaluation metric for regression problems. The idea of this metric is to be sensitive to relative errors. It is for example not changed by a global scaling of the target variable.
Evaluate Estimators -> Select Estimator -> Select Metrics -> Forecasting -> Roc
<PLACE HOLDER>
Evaluate Estimators -> Select Estimator -> Select Metrics -> Refit Score
Refit an estimator using the best-found parameters on the whole dataset. For multiple metric evaluation, this needs to be a str denoting the scorer that would be used to find the best parameters for refitting the estimator at the end.
Evaluate Estimators -> Select Estimator -> Select Metrics -> Iteration count for each model
Number of parameter settings that are sampled. n_iter trades off runtime vs quality of the solution.
Evaluate Estimators -> Select Estimator -> Select Metrics -> Number of folds in cross validation
Cross-validation is a technique for evaluating ML models by training several ML models on subsets of the available input data and evaluating them on the complementary subset of the data.
Evaluate Estimators -> Select Estimator -> Select Metrics -> Number of folds in cross validation -> Show Recommendation
Recommendation for number of Cross validation Folds.
Evaluate Estimators -> Select Estimator -> Select Algorithms
Selecting Different algorithms to train the ML model.
Evaluate Estimators -> Select Estimator -> Select Algorithms -> Logistic Regression
It is a statistical linear model for binary classification. The models uses the logistic function to calculate the probabilities for the target given an observation.
Evaluate Estimators -> Select Estimator -> Select Algorithms -> Decision Tree
This model predicts the value of a target variable by learning simple decision rules inferred from the data features.
Evaluate Estimators -> Select Estimator -> Select Algorithms -> XG Boost
This model make the prediction using an ensemble of weak prediction models, which are commonly decision trees.
Evaluate Estimators -> Select Estimator -> Select Algorithms -> Random Forest
Consists of training a large number of individual decision trees on various sub-samples of the dataset and uses averaging to improve the predictive accuracy and control over-fitting. The class with the most votes becomes our model’s prediction.
Evaluate Estimators -> Select Estimator -> Select Algorithms -> GaussianNB
Implements the Naive Bayes algorithm for classification. The likelihood of the features is assumed to be Gaussian. Naive Bayes methods are a set of supervised learning algorithms based on applying Bayes’ theorem with the “naive” assumption of conditional independence between every pair of features given the value of the class variable.
Evaluate Estimators -> Select Estimator -> Select Algorithms -> KNeighborsClassifier
Is a type of instance-based learning or non-generalizing learning: the input consists of the k closest training examples in a data set and the output is a class membership. An object is classified by a plurality vote of its neighbors, with the object being assigned to the class most common among its ‘k’ nearest neighbors (k is a positive integer).
Evaluate Estimators -> Select Estimator -> Select Algorithms -> Support Vector Classification
Is a supervised machine learning algorithm, we plot each data item as a point in n-dimensional space (where n is a number of features you have) with the value of each feature being the value of a particular coordinate. Then, we perform classification by finding the hyper-plane that differentiates the two classes well.
Evaluate Estimators -> Select Estimator -> Select Algorithms -> Linear Regression
Ordinary least squares Linear Regression fits a linear model with coefficients w = (w1, …, wp) to minimize the residual sum of squares between the observed targets in the dataset, and the targets predicted by the linear approximation.
Evaluate Estimators -> Select Estimator -> Select Algorithms -> Lasso Regression
Linear Model trained with L1 prior as regularizer (aka the Lasso). Technically the Lasso model is optimizing the same objective function as the Elastic Net with l1_ratio=1.0 (no L2 penalty).
Evaluate Estimators -> Select Estimator -> Select Algorithms -> Ridge Regression
This model solves a regression model where the loss function is the linear least squares function and regularization is given by the l2-norm. Also known as Ridge Regression or Tikhonov regularization.
Evaluate Estimators -> Select Estimator -> Select Algorithms -> Support Vector Regression (SVR)
Support Vector Regression (SVR) is a regression function that is generalized by Support Vector Machines. In this model a margin of tolerance (epsilon) is set in approximation to the SVM which would have already requested from the problem. The main idea is to minimize error, individualizing the hyperplane which maximizes the margin, keeping in mind that part of the error is tolerated.
Evaluate Estimators -> Select Estimator -> Select Algorithms -> random forest Regression
A random forest is a meta estimator that fits a number of regression decision trees on various sub-samples of the dataset and uses averaging to make more accurate predictions than any individual model.
Evaluate Estimators -> Select Estimator -> Select Algorithms -> Decision tree regression
Decision tree regression observes features of an object and trains a model in the structure of a tree to predict data in the future so it can produce meaningful continuous output. For example, a profit prediction model states the probable profit that can be generated from the sale of a product.
Evaluate Estimators -> Select Estimator -> Select Algorithms -> XGBoost regression
XGBoost is a decision-tree-based ensemble Machine Learning algorithm that uses a gradient boosting framework.
Evaluate Estimators -> Select Estimator -> Select Algorithms -> ExponentialSmoothing
Holt-Winters Exponential Smoothing is used for forecasting time series data that exhibits both a trend and a seasonal variation. This method comprises the forecast equation and three smoothing equations, one for the level, one for the trend and one for the seasonal component.
Evaluate Estimators -> Select Estimator -> Select Algorithms -> AUTO ARIMA
The auto-ARIMA algorithm seeks to identify the most optimal parameters for an ARIMA model. An ARIMA, or autoregressive integrated moving average model, is a generalization of an autoregressive moving average (ARMA) model, and is fitted to time-series data in an effort to forecast future points. ARIMA models can be especially efficacious in cases where data shows evidence of non-stationarity.
Evaluate Estimators -> Select Estimator -> Select Algorithms -> Polynomial Trend
Forecast time series data with a polynomial trend. The settings for this algorithm is train a linear regression model with a 1 or more degree polynomial transformation of the feature.
Evaluate Estimators -> Confirm Estimators
To select the estimator based on the metrics calculated.
Evaluate Estimators -> Confirm Estimators -> Accuracy Score
Accuracy is the fraction of predictions our model got right. Formally, accuracy has the following definition: Accuracy = Number of correct Predictions / Total Number of predictions. For binary classification, accuracy can also be calculated in terms of positives and negatives as follows: Accuracy = (TP+TN)/(TP+TN+FP+FN). Where TP = True Positives, TN = True Negatives, FP = False Positives, and FN = False Negatives.
Evaluate Estimators -> Confirm Estimators -> F1 Score
The F1 score is defined as the harmonic mean of precision and recall. In the F1 score, we compute the average of precision and recall. F1 Score = 2 * ((Precision * Recall)/ (Precision + Recall)). the F1 score gives equal weight to Precision and Recall:
Evaluate Estimators -> Confirm Estimators -> Recall Score
It is the proportion of actual positives that were identified correctly. Recall = TP / (TP+FN). Where TP = True Positives and FN = False Negatives.
Evaluate Estimators -> Confirm Estimators -> Precision Score
It is the proportion of positive identifications that were actually correct. Precision = TP / (TP+FP). Where TP = True Positives and FP = False Positives.
Evaluate Estimators -> Confirm Estimators -> Roc
A receiver operating characteristic curve, or ROC curve, is a graphical plot that illustrates the diagnostic ability of a binary classifier system as its discrimination threshold is varied.
Evaluate Estimators -> Confirm Estimators -> Confusion Matrix
A confusion matrix, also known as an error matrix, is a specific table layout that allows visualization of the performance of an algorithm, typically a supervised learning one (in unsupervised learning it is usually called a matching matrix). Each row of the matrix represents the instances in an actual class while each column represents the instances in a predicted class, or vice versa.
Tune Estimators
Hyperparameter optimization or tuning is the problem of choosing a set of optimal hyperparameters for a learning algorithm.
Tune Estimators -> Iterations Value
Number of parameter settings that are sampled. n_iter trades off runtime vs quality of the solution.
Tune Estimators -> Advance Tuning hyperparameter
Advance Hyperparameters to tune and find an optimal params.
Tune Estimators -> Advance Tuning Parameter Logistic Regression-> C
float, default=1.0. Inverse of regularization strength; must be a positive float. Like in support vector machines, smaller values specify stronger regularization.
Tune Estimators -> Advance Tuning Parameter Logistic Regression-> Fit intercept
bool, default=True. Specifies if a constant (a.k.a. bias or intercept) should be added to the decision function.
Tune Estimators -> Advance Tuning Parameter Logistic Regression-> Penalty
penalty {'l1', 'l2', 'none'}, default='l2'. Specify the norm of the penalty.'none': no penalty is added,'l2': add a L2 penalty term and it is the default choice.'elasticnet': both L1 and L2 penalty terms are added.
Tune Estimators -> Advance Tuning Parameter Logistic Regression-> Penalty --. L1 and L2
L1 adds the “absolute value of magnitude” of the coefficient as a penalty term to the loss function. L2 adds the “squared magnitude” of the coefficient as the penalty term to the loss function.
Tune Estimators -> Advance Tuning Parameter Logistic Regression-> Class Weight
dict or 'balanced', default=None. Weights associated with classes in the form {class_label: weight}. If not given, all classes are supposed to have weight one. The “balanced” mode uses the values of y to automatically adjust weights inversely proportional to class frequencies in the input data as n_samples / (n_classes * np.bincount(y)).
Tune Estimators -> Advance Tuning Parameter -> Decision tree -> Max Depth
bool, default=True. Specifies if a constant (a.k.a. bias or intercept) should be added to the decision function.
Tune Estimators -> Advance Tuning Parameter -> Decision tree -> Min samples split
The minimum number of samples required to split an internal node. If int, then consider min_samples_split as the minimum number. If float, then min_samples_split is a fraction and ceil(min_samples_split * n_samples) are the minimum number of samples for each split.
Tune Estimators -> Advance Tuning Parameter -> Decision tree -> Min samples leaf
int or float, default=1. The minimum number of samples required to be at a leaf node. A split point at any depth will only be considered if it leaves at least min_samples_leaf training samples in each of the left and right branches. This may have the effect of smoothing the model, especially in regression. If int, then consider min_samples_leaf as the minimum number. If float, then min_samples_leaf is a fraction and ceil(min_samples_leaf * n_samples) are the minimum number of samples for each node.
Tune Estimators -> Advance Tuning Parameter -> Decision tree -> Max features
int, float or {“auto”, “sqrt”, “log2”}, default=None. The number of features to consider when looking for the best split, If int, then consider max_features features at each split.
If float, then max_features is a fraction and int (max_features * n_features) features are considered at each split.
If “auto”, then max_features=sqrt(n_features).
If “sqrt”, then max_features=sqrt(n_features).
If “log2”, then max_features=log2(n_features).
If None, then max_features=n_features.
Tune Estimators -> Advance Tuning Parameter -> Decision tree -> class weight
dict, list of dict or “balanced”, default=None. Weights associated with classes in the form {class_label: weight}. If None, all classes are supposed to have weight one. For multi-output problems, a list of dicts can be provided in the same order as the columns of y. Note that for multioutput (including multilabel) weights should be defined for each class of every column in its own dict. For example, for four-class multilabel classification weights should be [{0: 1, 1: 1}, {0: 1, 1: 5}, {0: 1, 1: 1}, {0: 1, 1: 1}] instead of [{1:1}, {2:5}, {3:1}, {4:1}]. The “balanced” mode uses the values of y to automatically adjust weights inversely proportional to class frequencies in the input data as n_samples / (n_classes * np.bincount(y)). For multi-output, the weights of each column of y will be multiplied.
Tune Estimators -> Advance Tuning Parameter -> XG Boost -> learning_rate
range: [0,1], default=0.3. Step size shrinkage used in update to prevent overfitting. After each boosting step, we can directly get the weights of new features, and learning_rate shrinks the feature weights to make the boosting process more conservative.
Tune Estimators -> Advance Tuning Parameter -> XG Boost -> gamma
default=0. Minimum loss reduction required to make a further partition on a leaf node of the tree. The larger gamma is, the more conservative the algorithm will be.
Tune Estimators -> Advance Tuning Parameter -> XG Boost -> Maximum depth
default=6. Maximum depth of a tree. Increasing this value will make the model more complex and more likely to overfit. 0 is only accepted in lossguide growing policy when tree_method is set as hist or gpu_hist and it indicates no limit on depth. Beware that XGBoost aggressively consumes memory when training a deep tree.
Tune Estimators -> Advance Tuning Parameter -> XG Boost -> col sample by tree
default=1. Is the subsample ratio of columns when constructing each tree. Subsampling occurs once every tree is constructed.
Tune Estimators -> Advance Tuning Parameter -> XG Boost -> sub sample ratio
range: (0,1], default=1. Subsample ratio of the training instances. Setting it to 0.5 means that XGBoost would randomly sample half of the training data prior to growing trees. and this will prevent overfitting. Subsampling will occur once in every boosting iteration.
Tune Estimators -> Advance Tuning Parameter -> XG Boost -> Reg_alpha
default=0. L1 regularization term on weights. Increasing this value will make the model more conservative.
Tune Estimators -> Advance Tuning Parameter -> XG Boost -> reg_lambda
default=1. L2 regularization term on weights. Increasing this value will make the model more conservative.
Tune Estimators -> Advance Tuning Parameter -> XG Boost -> min child weight
default=1. The minimum sum of instance weight (hessian) needed in a child. If the tree partition step results in a leaf node with the sum of instance weight less than min_child_weight, then the building process will give up further partitioning. In a linear regression task, this simply corresponds to a minimum number of instances needed to be in each node. The larger min_child_weight is, the more conservative the algorithm will be.
Tune Estimators -> Advance Tuning Parameter -> XG Boost -> n_estimators
The number of runs XGBoost will try to learn
Tune Estimators -> Advance Tuning Parameter -> Random forest -> Max Depth
bool, default=True. Specifies if a constant (a.k.a. bias or intercept) should be added to the decision function.
Tune Estimators -> Advance Tuning Parameter -> Random forest -> Min samples split
The minimum number of samples required to split an internal node. If int, then consider min_samples_split as the minimum number. If float, then min_samples_split is a fraction and ceil(min_samples_split * n_samples) are the minimum number of samples for each split.
Tune Estimators -> Advance Tuning Parameter -> Random forest -> Min samples leaf
int or float, default=1. The minimum number of samples required to be at a leaf node. A split point at any depth will only be considered if it leaves at least min_samples_leaf training samples in each of the left and right branches. This may have the effect of smoothing the model, especially in regression. If int, then consider min_samples_leaf as the minimum number. If float, then min_samples_leaf is a fraction and ceil(min_samples_leaf * n_samples) are the minimum number of samples for each node.
Tune Estimators -> Advance Tuning Parameter -> Random forest -> Max features
int, float or {“auto”, “sqrt”, “log2”}, default=None. The number of features to consider when looking for the best split , if int, then consider max_features features at each split.
If float, then max_features is a fraction, and int (max_features * n_features) features are considered at each split.
If “auto”, then max_features=sqrt(n_features).
If “sqrt”, then max_features=sqrt(n_features).
If “log2”, then max_features=log2(n_features).
If None, then max_features=n_features.
Tune Estimators -> Advance Tuning Parameter -> Random forest -> class weight
dict, list of dict or “balanced”, default=None. Weights associated with classes in the form {class_label: weight}. If None, all classes are supposed to have weight one. For multi-output problems, a list of dicts can be provided in the same order as the columns of y. Note that for multioutput (including multilabel) weights should be defined for each class of every column in its own dict. For example, for four-class multilabel classification weights should be [{0: 1, 1: 1}, {0: 1, 1: 5}, {0: 1, 1: 1}, {0: 1, 1: 1}] instead of [{1:1}, {2:5}, {3:1}, {4:1}]. The “balanced” mode uses the values of y to automatically adjust weights inversely proportional to class frequencies in the input data as n_samples / (n_classes * np.bincount(y)). For multi-output, the weights of each column of y will be multiplied.
Tune Estimators -> Advance Tuning Parameter -> Random forest -> n estimators
The number of runs random forest will try to learn.
Tune Estimators -> Advance Tuning Parameter -> Gaussian NB -> var_smoothing
Portion of the largest variance of all features that is added to variances for calculation stability.
Tune Estimators -> Advance Tuning Parameter -> KNeighborsClassifier -> n_neighbors
Number of neighbors to use by default for kneighbors queries.
Tune Estimators -> Advance Tuning Parameter -> KNeighborsClassifier -> weights
Weight function used in prediction. Possible values:
‘uniform’ : uniform weights. All points in each neighborhood are weighted equally.
‘distance’ : weight points by the inverse of their distance. in this case, closer neighbors of a query point will have a greater influence than neighbors which are further away.
[callable] : a user-defined function which accepts an array of distances and returns an array of the same shape containing the weights.
Tune Estimators -> Advance Tuning Parameter -> KNeighborsClassifier -> p
Power parameter for the Minkowski metric. When p = 1, this is equivalent to using manhattan_distance (l1), and euclidean_distance (l2) for p = 2. For arbitrary p, minkowski_distance (l_p) is used.
Tune Estimators -> Advance Tuning Parameter -> KNeighborsClassifier -> Leaf size
Leaf size passed to BallTree or KDTree. This can affect the speed of the construction and query, as well as the memory required to store the tree. The optimal value depends on the nature of the problem.
Tune Estimators -> Advance Tuning Parameter -> SVC -> C
Regularization parameter. The strength of the regularization is inversely proportional to C. Must be strictly positive. The penalty is a squared l2 penalty.
Tune Estimators -> Advance Tuning Parameter -> SVC -> kernel
{‘linear’, ‘poly’, ‘rbf’, ‘sigmoid’, ‘precomputed’} or callable, default=’rbf’
Specifies the kernel type to be used in the algorithm. If none is given, ‘rbf’ will be used. If a callable is given it is used to pre-compute the kernel matrix from data matrices; that matrix should be an array of shape (n_samples, n_samples).
Tune Estimators -> Advance Tuning Parameter -> SVC -> degree
Degree of the polynomial kernel function (‘poly’). Must be non-negative. Ignored by all other kernels.
Tune Estimators -> Advance Tuning Parameter -> SVC -> gamma
{‘scale’, ‘auto’} or float, default=’scale’
Kernel coefficient for ‘rbf’, ‘poly’ and ‘sigmoid’.
if gamma='scale' (default) is passed then it uses 1 / (n_features * X.var()) as value of gamma,
if ‘auto’, uses 1 / n_features
if float, must be non-negative.
Tune Estimators -> Advance Tuning Parameter -> Linear regression -> fit intercept
bool, default=True.Whether to calculate the intercept for this model. If set to False, no intercept will be used in calculations (i.e. data is expected to be centered).
Tune Estimators -> Advance Tuning Parameter -> lasso regression -> Alpha
Constant that multiplies the L1 term, controlling regularization strength. alpha must be a non-negative float i.e. in [0, inf).
When alpha = 0, the objective is equivalent to ordinary least squares, solved by the LinearRegression object. For numerical reasons, using alpha = 0 with the Lasso object is not advised. Instead, you should use the LinearRegression object.
Tune Estimators -> Advance Tuning Parameter -> Lasso regression -> fit intercept
Whether to calculate the intercept for this model. If set to False, no intercept will be used in calculations (i.e. data is expected to be centered).
Tune Estimators -> Advance Tuning Parameter -> Ridge regression -> Alpha
Constant that multiplies the L2 term, controlling regularization strength. alpha must be a non-negative float i.e. in [0, inf).
When alpha = 0, the objective is equivalent to ordinary least squares, solved by the LinearRegression object. For numerical reasons, using alpha = 0 with the Ridge object is not advised. Instead, you should use the LinearRegression object.
If an array is passed, penalties are assumed to be specific to the targets. Hence they must correspond in number.
Tune Estimators -> Advance Tuning Parameter -> Ridge regression -> fit intercept
Whether to calculate the intercept for this model. If set to False, no intercept will be used in calculations (i.e. data is expected to be centered).
Tune Estimators -> Advance Tuning Parameter -> Support Vector regression -> C
float, default=1.0
Regularization parameter. The strength of the regularization is inversely proportional to C. Must be strictly positive. The penalty is a squared l2 penalty.
Tune Estimators -> Advance Tuning Parameter -> Support Vector regression -> kernel
{‘linear’, ‘poly’, ‘rbf’, ‘sigmoid’, ‘precomputed’} or callable, default=’rbf’
Specifies the kernel type to be used in the algorithm. If none is given, ‘rbf’ will be used. If a callable is given it is used to precompute the kernel matrix.
Tune Estimators -> Advance Tuning Parameter -> Support Vector regression -> gamma
{‘scale’, ‘auto’} or float, default=’scale’
Kernel coefficient for ‘rbf’, ‘poly’ and ‘sigmoid’.
if gamma='scale' (default) is passed then it uses 1 / (n_features * X.var()) as value of gamma,
if ‘auto’, uses 1 / n_features
if float, must be non-negative.
Tune Estimators -> Advance Tuning Parameter -> Support Vector regression -> degree
int, default=3
Degree of the polynomial kernel function (‘poly’). Must be non-negative. Ignored by all other kernels.
Tune Estimators -> Advance Tuning Parameter -> RandomForestRegressor -> n_estimators
The number of trees in the forest.
Tune Estimators -> Advance Tuning Parameter -> RandomForestRegressor -> max_depth
The maximum depth of the tree. If None, then nodes are expanded until all leaves are pure or until all leaves contain less than min_samples_split samples.
Tune Estimators -> Advance Tuning Parameter -> RandomForestRegressor -> criterion
{“squared_error”, “absolute_error”, “friedman_mse”, “poisson”}, default=”squared_error”
The function to measure the quality of a split. Supported criteria are “squared_error” for the mean squared error, which is equal to variance reduction as feature selection criterion and minimizes the L2 loss using the mean of each terminal node, “friedman_mse”, which uses mean squared error with Friedman’s improvement score for potential splits, “absolute_error” for the mean absolute error, which minimizes the L1 loss using the median of each terminal node, and “poisson” which uses reduction in Poisson deviance to find splits. Training using “absolute_error” is significantly slower than when using “squared_error”.
Tune Estimators -> Advance Tuning Parameter -> RandomForestRegressor -> min_samples_split
The minimum number of samples required to split an internal node:
If int, then consider min_samples_split as the minimum number.
If float, then min_samples_split is a fraction and ceil(min_samples_split * n_samples) are the minimum number of samples for each split.
Tune Estimators -> Advance Tuning Parameter -> RandomForestRegressor -> min_samples_leaf
The minimum number of samples required to be at a leaf node. A split point at any depth will only be considered if it leaves at least min_samples_leaf training samples in each of the left and right branches. This may have the effect of smoothing the model, especially in regression.
If int, then consider min_samples_leaf as the minimum number.
If float, then min_samples_leaf is a fraction and ceil(min_samples_leaf * n_samples) are the minimum number of samples for each node.
Tune Estimators -> Advance Tuning Parameter -> RandomForestRegressor -> max_features
The number of features to consider when looking for the best split:
If int, then consider max_features features at each split.
If float, then max_features is a fraction and max(1, int(max_features * n_features_in_)) features are considered at each split.
If “auto”, then max_features=n_features.
If “sqrt”, then max_features=sqrt(n_features).
If “log2”, then max_features=log2(n_features).
If None or 1.0, then max_features=n_features.
Tune Estimators -> Advance Tuning Parameter -> DecisionTreeRegressor -> criterion
{“squared_error”, “friedman_mse”, “absolute_error”, “poisson”}, default=”squared_error”
The function to measure the quality of a split. Supported criteria are “squared_error” for the mean squared error, which is equal to variance reduction as feature selection criterion and minimizes the L2 loss using the mean of each terminal node, “friedman_mse”, which uses mean squared error with Friedman’s improvement score for potential splits, “absolute_error” for the mean absolute error, which minimizes the L1 loss using the median of each terminal node, and “poisson” which uses reduction in Poisson deviance to find splits.
Tune Estimators -> Advance Tuning Parameter -> DecisionTreeRegressor -> max_depth
The maximum depth of the tree. If None, then nodes are expanded until all leaves are pure or until all leaves contain less than min_samples_split samples.
Tune Estimators -> Advance Tuning Parameter -> DecisionTreeRegressor -> min_samples_split
int or float, default=2
The minimum number of samples required to split an internal node:
If int, then consider min_samples_split as the minimum number.
If float, then min_samples_split is a fraction and ceil(min_samples_split * n_samples) are the minimum number of samples for each split.
Tune Estimators -> Advance Tuning Parameter -> DecisionTreeRegressor -> min_samples_leaf
int or float, default=1
The minimum number of samples required to be at a leaf node. A split point at any depth will only be considered if it leaves at least min_samples_leaf training samples in each of the left and right branches. This may have the effect of smoothing the model, especially in regression.
If int, then consider min_samples_leaf as the minimum number.
If float, then min_samples_leaf is a fraction and ceil(min_samples_leaf * n_samples) are the minimum number of samples for each node.
Tune Estimators -> Advance Tuning Parameter -> DecisionTreeRegressor -> max_features
int, float or {“auto”, “sqrt”, “log2”}, default=None
The number of features to consider when looking for the best split:
If int, then consider max_features features at each split.
If float, then max_features is a fraction and max(1, int(max_features * n_features_in_)) features are considered at each split.
If “auto”, then max_features=n_features.
If “sqrt”, then max_features=sqrt(n_features).
If “log2”, then max_features=log2(n_features).
If None, then max_features=n_features.
Tune Estimators -> Advance Tuning Parameter -> XGB Regressor -> learning_rate
float, default=0.1
Learning rate shrinks the contribution of each tree by learning_rate. There is a trade-off between learning_rate and n_estimators. Values must be in the range [0.0, inf).
Tune Estimators -> Advance Tuning Parameter -> XGB Regressor -> gamma
[default=0, alias: min_split_loss]
Minimum loss reduction required to make a further partition on a leaf node of the tree. The larger gamma is, the more conservative the algorithm will be.
range: [0,∞]
Tune Estimators -> Advance Tuning Parameter -> XGB Regressor -> col sample by tree
is the subsample ratio of columns when constructing each tree. Subsampling occurs once for every tree constructed.
Tune Estimators -> Advance Tuning Parameter -> XGB Regressor -> subsample
[default=1]
Subsample ratio of the training instances. Setting it to 0.5 means that XGBoost would randomly sample half of the training data prior to growing trees. and this will prevent overfitting. Subsampling will occur once in every boosting iteration.
range: (0,1]
Tune Estimators -> Advance Tuning Parameter -> XGB Regressor -> reg_alpha
[default=0, alias: reg_alpha]
L1 regularization term on weights. Increasing this value will make model more conservative. Normalised to number of training examples.
Tune Estimators -> Advance Tuning Parameter -> XGB Regressor -> reg_lambda
[default=1, alias: reg_lambda]
L2 regularization term on weights. Increasing this value will make model more conservative.
Tune Estimators -> Advance Tuning Parameter -> XGB Regressor -> min_child_weight
[default=1]
Minimum sum of instance weight (hessian) needed in a child. If the tree partition step results in a leaf node with the sum of instance weight less than min_child_weight, then the building process will give up further partitioning. In linear regression task, this simply corresponds to minimum number of instances needed to be in each node. The larger min_child_weight is, the more conservative the algorithm will be.
range: [0,∞]
Tune Estimators -> Advance Tuning Parameter -> XGB Regressor -> n_estimators
The number of trees in the forest.
Tune Estimators -> Advance Tuning Parameter -> ExponentialSmoothing -> trend
{"add", "mul", "additive", "multiplicative", None}, default=None Type of trend component.
Tune Estimators -> Advance Tuning Parameter -> ExponentialSmoothing -> damped_trend
bool, default=False. Should the trend component be damped.
Tune Estimators -> Advance Tuning Parameter -> ExponentialSmoothing -> seasonal
{"add", "mul", "additive", "multiplicative", None}, default=None
Type of seasonal component.Takes one of
Tune Estimators -> Advance Tuning Parameter -> ExponentialSmoothing -> sp
int or None, default=None The number of seasonal periods to consider.
Tune Estimators -> Advance Tuning Parameter -> ExponentialSmoothing -> use_boxcox
{True, False, 'log', float}, default=None Should the Box-Cox transform be applied to the data first? If 'log' then apply the log. If float then use lambda equal to float.
Tune Estimators -> Advance Tuning Parameter -> AUTO ARIMA -> seasonal
bool, optional (default=True) Whether to fit a seasonal ARIMA. Default is True. Note that if seasonal is True and sp == 1, seasonal will be set to False.
Tune Estimators -> Advance Tuning Parameter -> AUTO ARIMA -> stationary
bool, optional (default=False) Whether the time-series is stationery and d should be set to zero.
Tune Estimators -> Advance Tuning Parameter -> AUTO ARIMA -> information_criterion
str, optional (default='aic') The information criterion used to select the best ARIMA model. One of pmdarima.arima.auto_arima.VALID_CRITERIA, (‘aic’, ‘bic’, ‘hqic’, ‘oob’).
Tune Estimators -> Advance Tuning Parameter -> AUTO ARIMA -> trend
str, optional (default=None) The trend parameter. If with_intercept is True, trend will be used. If with_intercept is False, the trend will be set to a no- intercept value.
Tune Estimators -> Advance Tuning Parameter -> Polynomial Trend -> degree
int, default = 1 Degree of polynomial function
Tune Estimators -> Advance Tuning Parameter -> Polynomial Trend -> with_intercept
bool, default=True If true, then include a feature in which all polynomial powers are zero. (i.e. a column of ones, acts as an intercept term in a linear model)