Way 1 :

• Delete all records that were registered before creating the store.
• OneHotEncoder: date, holidays
• Label Encoder: family , city ,state , type
• Using average window 6 —> fill nulls and edit all recored
• Create_lags
• Create_rolling_mean
• Exponential moving average of target (sales)
• GridSearch
• Except all products that have been discontinued

Way 2 (Dart)

• Convert data to time series : using TimeSeries in Darts
• Scaler(Standard Scaler) : refers to a data transformation technique used to scale or normalize the values of a time series. Scaling is a common preprocessing step in time series analysis and forecasting to ensure that different time series have a similar scale or magnitude.
• StaticCovariatesTransformer: used to preprocess static covariate data associated with time series. Static covariates are additional variables or features that do not change over time and provide additional information that can help improve the modeling and forecasting of time series data.
• MissingValuesFiller: used to fill missing values in time series data. Missing values are gaps or missing observations in a time series that can occur due to various reasons such as data collection errors, sensor failures, or data processing issues.
• InvertibleMapper : is a concept that refers to a mapping or transformation that can be reversed or inverted. It is commonly used in data preprocessing or feature engineering pipelines to apply a specific transformation to the data and then recover the original data if needed.
• MovingAverage: used statistical technique for smoothing time series data by calculating the average of a specific window or interval of consecutive data points. It helps to reduce noise and highlight the underlying trend or pattern in the data.