> For the complete documentation index, see [llms.txt](https://doc.verteego.com/verteego-doc/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://doc.verteego.com/verteego-doc/pipelines/forecasting-pipelines/calculators/mathematic/moving-average-ewm.md).

# Moving Average (EWM)

Description of the EWM calculator that allow to compute moving average.

The **Exponential Weighted Function** (EWM) calculator is a tool within the Verteego platform that allows users to apply exponential smoothing to a dataset for various statistical operations. The calculator supports multiple methods including **mean**, **variance (var)**, **standard deviation (std)**, **correlation (corr)**, and **covariance (cov)**.

**Input Columns**

These are the columns used in the calculation for the exponential weighted function.

* **Names of Columns**: Specify the columns that will be used to apply the calculation.

**Output Columns**

These are the columns that will be added to the dataset as a result of the calculation.

* **Names of Columns**: The output columns generated by applying the exponential weighted function to the input columns.

**Parameters**

This section defines the parameters required for the exponential weighted function calculation.

1. **method**:
   * This specifies the method used for smoothing the values in the input columns. The supported methods are:
     * `mean`: Computes the exponential weighted mean.
     * `var`: Computes the exponential weighted variance.
     * `std`: Computes the exponential weighted standard deviation.
     * `corr`: Computes the exponential weighted correlation.
     * `cov`: Computes the exponential weighted covariance.
2. **span**:
   * Defines the span of the smoothing window, typically in terms of the number of rows considered for the exponential weighting. A larger span results in smoother values, while a smaller span gives more weight to recent values.
3. **values**:
   * A list of columns that need to be smoothed. These columns will undergo the exponential weighted smoothing calculation.
4. **sort\_by**:
   * A list of column names by which the dataset should be sorted before the calculation. Sorting ensures the correct chronological or logical order of the data before applying the smoothing.
5. **group\_by**:
   * A list of columns used for grouping the data. The smoothing function will be applied to each group individually.

***

**Example in YAML Format**

This example demonstrates how to configure the exponential weighted function in YAML format for a typical use case:

```yaml
yamlCopyEditcalculated_cols:
  qty_sold_ewm:
    method: ewm
    input_columns:
      - qty_sold
      - itemid
      - locid
      - date
    output_columns:
      - qty_sold_ewm
    params:
      span: 4
      method: mean
      values:
        - qty_sold
      sort_by:
        - itemid
        - locid
        - date
      group_by:
        - itemid
        - locid
```

In this YAML configuration:

* The exponential weighted function (`ewm`) is applied to the `qty_sold` column.
* The smoothing method used is **mean** (`method: mean`).
* The smoothing window spans over 4 rows (`span: 4`).
* The dataset will be sorted by `itemid`, `locid`, and `date` before applying the smoothing (`sort_by: [itemid, locid, date]`).
* The smoothing will be applied to the `qty_sold` column and grouped by `itemid` and `locid` (`group_by: [itemid, locid]`).
* The output will be stored in the `qty_sold_ewm` column.

***

#### **Use Cases**

This calculator can be useful in various analytical scenarios where smoothing is needed to make time series data more interpretable or to detect trends. Some use cases include:

* **Sales forecasting**: Apply exponential weighted mean or variance to sales data to reduce noise and better capture trends.
* **Demand forecasting**: Smooth the demand data to predict future stock levels or adjust replenishment schedules.
* **Financial analysis**: Use correlation or covariance for financial data to understand the relationships between different assets or business factors over time.

***

#### **Benefits**

* **Real-time Data Smoothing**: Automatically smooth your data to reveal clearer patterns.
* **Enhanced Accuracy**: By choosing the right method (mean, var, std, etc.), you can improve decision-making based on the smoothed values.
* **Customizable for Different Data Types**: Supports a variety of use cases by adjusting parameters like span, sorting, and grouping.