Calculators

Calculators are designed to dynamically enhance your datasets by calculating new variables that significantly boost the performance of your forecasting models. These calculators provide a powerful means to enrich your data, ensuring more robust and insightful predictive analytics.

Types of Calculators

Verteego offers five distinct types of calculators, each tailored to generate additional columns based on specific needs:

External

These calculators augment your datasets with columns joined from external sources. For example, they can integrate data from our weather API or other imported datasets, providing contextual information that enhances model predictions.

• get_from_dataset

• weather

Mathematic

Mathematic calculators create new columns by applying various mathematical methods to existing data. These operations can range from simple binary operations to complex mathematical expressions, enabling detailed and customized data manipulations.

• aggregate_val_group_by_key

• binary_operation

• count_rows_by_keys

• hierarchical_aggregate

• mathematical_expression

• unary_operation

Machine Learning

Utilize machine learning algorithms to generate sophisticated data columns. Examples include creating word similarity scores, applying one-hot encoding, or using generalized linear mixed models (GLMM) encoding. These techniques transform raw data into formats that are more amenable to advanced analytical models.

• pca

• clustering

• glmm_encoder

• one_hot_encode

• words_similarity

Transformation

Transformation calculators are used to modify existing data columns, improving data quality or adjusting data formats. Common operations include filling missing values (NA), replacing text strings, or other value transformations that standardize data inputs for more consistent analysis.

• fillna

• case_na

• interval_index

• constant

• cyclic

• replace

Temporal

Focus on extracting and creating temporal features from date columns. These calculators can identify holidays, extract date attributes (such as day of the week or month), assess seasonality, and mark special events. Temporal features are crucial for models that rely on time-based patterns and seasonal trends.

• bank_holidays_countdown

• bankholidays

• date_attributes

• date_weight

• day_count

• duration

• events_countdown

• seasonality

• tsfresh