Managing Logs

Overview

Process Mining is a fully featured Object-Centric Process Mining platform. It allows the merging and reorganization of logs using multiple attributes as case IDs, as well as association of a single event with multiple objects. The platform has numerous visualization options to view data from different perspectives so that users can draw the best insights from them.

Logs work best with Event driven data - essentially data that has a time stamp and can be tracked.

Importing Logs

Like any other file, Logs are imported from the Import/Export menu at the top of the page.

If importing a CSV file, select NO when asked if it is to be used to train a predictive model.

Attributes will need to be set. Case ID and Activity must be set to proceed. For more on the import process see Importing/Exporting Process Mining Data.

The imported Log will have an icon like this:

You can right-click on the Log to get more details about it.

Merging Logs

To merge multiple Logs together, right click a Log, and select Merge. Then pick ADD LOGS to pick the Log or Logs that will be merged with the original. The Attributes that will make up the merged Log can be chosen by clicking on them. By default all Attributes are selected.

After pressing NEXT, Attribute labels will need to be paired. This can be done manually or users can have Process Mining automate it by pressing the magic wand icon in the middle of the screen.After pressing NEXT, users will be taken to a screen similar to the Import Log screen where Attributes will need to be labeled.

Columns with red headers will be ignored. To import a column drag and drop a tag from this panel to the desired column, or alternatively right-click on the column and select the desired tag. To remove a tag double-click on it or drag it back to this panel.

After that is complete, choose where the merged Log should be saved.

Once the Merge Log is saved users have the opportunity to set the order data is sorted by. Then users will be taken to the Analyzer screen and can begin visualizing their data. See Process Map Analyzer Overview to learn more about how to view and use the data.

Log Best Practices

Hot Data VS Cold Data

For each event log, users may set the data timeframe to keep in memory for faster analysis.

The data is divided into hot and cold data. Hot data is defined as the most recent data that is analyzed frequently. Hot data is stored in memory to allow faster analysis. On the other hand, cold data is defined as the old data that is not analyzed frequently. Cold data is stored on disk, which may lead to a slower analysis.

Dividing data into hot and cold allows faster analysis of the most recent and interesting data in memory without running into memory shortage issues, caused by loading all data in memory.

Setting Hot Data Window Size

Users may set the hot data window size to unlimited to load and analyze all data in memory. This is suitable for smaller event logs. Alternatively, users may set the hot data window size to a fixed number of months or years, in which case the most recent data that falls into the hot data window is loaded in memory. For example, if the hot data window size is set to six months, the cases that have started within the last six months are considered as hot data and loaded in memory.

It is recommended to set hot data window size such that it is large enough to keep the most frequently analyzed data in memory to speed up the user's regular analysis. An estimate of the memory usage of hot data is provided to assist you with setting the hot data window size and with avoiding running out of memory.

Default VS Custom Timeframe For Analysis

When opening an event log, by default, only the data that falls within the hot data window is included in the analysis, provided that the hot data window is not set to zero month/year. This is to allow faster opening of an event log, as well as faster analysis on the most recent and interesting data. On the other hand, if the hot data window is set to zero month/year, by default, all (cold) data will be included in the analysis, which may lead to slower opening of an event log and slower analysis.

After opening an event log, each user may select their own custom data timeframe if they need to perform their analysis on data that is not included in the default timeframe. Their selected data timeframe can include hot data, cold data or a combination of both. For example, if hot data window is set to six months, by default, only the cases that fall in the hot data window (which are cases that started in the last six months) are included in the analysis. Now, if a user want to perform their analysis on the data from the last twelve months, after the log is opened, they can manually specify a custom data timeframe of last twelve months (that includes six months of cold data and six months of hot data).