Batches

The functionality of batch processing (batches) allows you to perform calculations for large amounts of data. Batches are set of commands and algorithms that effectively process many items at a time (usually all products in the chain). There are also technical batches that allow you to delete the file cache, train the neural network, etc. Usually, batches calculations take a lot of time and they can be launched either by necessity or according to a schedule (as for example for forecasting and price monitoring batches).

Users are not advised to launch batch calculations before reading this section. A detailed description of each batch is given below:

Log - shows the full log of the system working process on the server.

Memory Report

Performance test - this batch launches a set of tests that evaluate server performance. So the system records the execution time of each test and compares with average performance times. This batch allows you to evaluate whether there is enough calculating power in the server on which MySales is running.

Forecasting batch - this batch launches a full recalculation of forecasts for all products. This batch takes a very long time to complete (from 4 to 10 hours, depending on the number of SKUs and stores). Forecasting batch is launched automatically at night according to a schedule agreed with the client to recalculate forecasts and transfer relevant data to the customers auto-order system. Manually, this batch is usually launched by the support service in case of errors during its scheduled calculation. Users can also run this batch if necessary.

Price Monitoring - this batch launches a competitors price update for those SKUs that were entered into the price monitoring module. Price monitoring batch starts automatically according to an agreed schedule.

Catalogues Planned Scrapping - this batch starts updating competitors catalogues for those catalogues that were entered into the catalogues monitoring module. Competitors catalogues update batch starts automatically according to an agreed schedule.

Catalogues Bulk Sku Mapping - This batch binds SKUs in your chain with products from competitor catalogues that were obtained during catalogue monitoring.

Train Promo Net - this batch launches neural network training to predict promos using latest promo and sales data. The neural network training batch starts automatically once a week according to an agreed schedule. Manually launching this batch must be agreed with the promo team and the responsible MySales employee.

Refresh Promo Match Data - this batch launches a search algorithm for comparable (similar) promos. This is an important batch, as new promo forecasts are based on data from comparable historical promos. If necessary, it allows you to manually start the search for comparable promos, in normal mode the system automatically searches for comparable promos when building a new promo forecast.

Full Promo Calculation - this batch launches a complete recalculation of forecasts for all promos in history. The working time of this batch, depending on the number of promos, can reach several days. This batch is launched by MySales employees and only in cases where the promo forecasting logic is improved to obtain more accurate results.

-14 Days Promo Calculation - this batch recalculates the promo forecast for all promos no older than 14 days, as well as for all future promos. This batch is launched once a week on an agreed schedule. Manually, this batch is usually launched by the support service in case of errors during its scheduled calculation. Users can also run this batch if necessary.

-200 Days Promo Calculation - this batch recalculates the promo forecast for all promos no older than 200 days, as well as for all future promos. This batch is launched as needed.

УTruncate All Forecasts - running this batch deletes all forecasts from the database. Users mustn't run it. It can only be run by a MySales employee, and only in certain situations.

Clear Cache Files - running this batch will delete all locally cached historical data. When building a forecast, MySales caches the historical data, so that next time forecast is build, system doesn't have to load again all the data from the database. This approach allows you to speed up the building of new forecasts and reduce the load on the database. If you execute this batch, then at the next recalculation of forecasts MySales will load again all historical data from the database, which will significantly increase the recalculation time. This batch needs to be started if there were changes in historical data.

Clear Saved Promo Group Net -

Collect Statistics - this batch is similar to the forecasting batch. It starts the calculation of forecasts for all SKUs in the network, but it does not save the calculated forecasts into the database, but only collects statistical information, which can be seen in the "Statistics" section. This allows you to collect data about forecast accuracy in a short time, since the data will not be written to the database (writing to the database is a very long process).

Calc Price Cross Elasticity - this batch calculates price cross elasticity for all SKUs. The calculation results can be seen in the "Price Optimization" section. This batch is launched once a week on an agreed schedule.

Load Sales Cache - this batch downloads all the necessary historical data from database for all SKUs to speed up the forecasting process. This batch starts reverse process of the "Clear Cache Files" batch. It makes sense to run it after running "Clear Cache Files" batch.

Refresh Sales By Groups - this batch recalculates group sales using SKU sales. It aggregates SKU sales into group sales. This batch is launched once a week on an agreed schedule. Manually, it makes sense to run it if the group sales and total SKU sales are different.