Algorithms grouped by purpose

Use clustering algorithms to assign each data point to a group (cluster) based on the similarities with other data points. For some algorithms, you need to specify the number of clusters to be found as a parameter. Other algorithms detect the number of clusters themselves.

The following algorithms are available:

• DCPY.AFFINITYPROPCLUST(max_iter, convergence_iter, damping, preference, columns)
• DCPY.AGGLOCLUST(n_clusters, affinity, linkage, columns)
• DCPY.AUTOCLUST(max_clusters, columns)
• DCPY.BIRCHCLUST(n_clusters, threshold, branching_factor, columns)
• DCPY.DBSCANCLUST(min_samples, eps, columns)
• DCPY.KMEANSCLUST(n_clusters, random_state, init, n_init, max_iter, columns)
• DCPY.MEANSHIFTCLUST(cluster_all, bandwidth, columns)
• DCPY.MINIBATCHKMEANSCLUST(n_clusters, random_state, init, n_init, batch_size, max_no_improvement, max_iter, columns)
• DCPY.SPECTRALCLUST(n_clusters, random_state, n_init, columns)
• MLLIB.BICLUSTER(imputer, n_clusters, seed, columns)
• MLLIB.CLUSTER(imputer, n_clusters, n_iter, columns)
• MLLIB.GMMCLUSTER(imputer, n_clusters, columns)

Use correlation algorithms to decide whether there is any correlation between the data from two columns. The resulted correlation coefficient can have values between -1 and 1. A value closer to 1 indicates a strong positive linear correlation. A value around 0 indicates no correlation. A value close to -1 indicates a strong negative linear correlation.

The following algorithms are available:

• MLLIB.PEARSONCOR(columns)
• MLLIB.SPEARMANCOR(columns)

Use the algorithms for outliers detection to identify the data points that deviate from the overall pattern of your data points. The algorithm assigns a value to the outlier data points and another value to the inlier data points. For example, the DCPY.LOFOUTLIER algorithm assigns the value '-1' to outliers and the value '1' to inliers.

The following algorithms are available:

• DCPY.LOFOUTLIER(n_neighbors, leaf_size, contamination, columns)
• DCPY.ENVELOPE(contamination, columns)
• DCPY.ISOLATIONFOREST(n_estimators, max_samples, max_features, contamination, columns)
• SMILE.OUTLIERS(std_deviation_X, columns)

Use regression algorithms to predict the values of the dependent variable from independent variables.

The MLLIB.LINEARREG(fit_intercept, standardize, columns) algorithm is available.

Use scaling algorithms to reduce the data values to a specific range using a mathematical transformation. For example, the DCPY.MAXABSSCALER algorithm reduces the absolute data values to the [0, 1] range by dividing each value by the largest absolute value in the column.

The following algorithms are available:

• DCPY.MAXABSSCALER(column)
• DCPY.MINMAXSCALER(feature_range_min, feature_range_max, column)
• DCPY.ROBUSTSCALER(with_centering, with_scaling, quantile_range_min, quantile_range_max, column)
• DCPY.STANDARDSCALER(with_mean, with_std, column)

Use transformation algorithms to apply various transformations to your data, such as shrinking the distances between marginal outliers and inliers (DCPY.QUANTILETRANSFORM) or imputing missing values in a column based on a specific strategy (DCPY.IMPUTER).

The following algorithms are available:

• DCPY.IMPUTER(strategy, column(s))
• DCPY.QUANTILETRANSFORM(n_quantiles, output_distribution, column)