In many analytics use cases, event data often plays a key role. This data may be a user’s interaction with an app on mobile or their interaction with a web application. This data is very valuable to understand the series of events a user has traversed in a given period of time.
One of the analyses we can do on this data is user session analysis. A session is a period of time the user comes to an application, spends some time in an application and moves away. This session analysis often helps to understand the various activities users will be doing in a short period of time.
In this series of blogs, we will discuss how Tinace helps users to perform session analysis on their event data. This is first part in which we will discuss about the use case and how to start with SQL in Tinace Platform
Sample Data
The above data shows the sample data of activities. Here event_at signifies the time at which an event has occurred. Other fields signify the user and activity information.
As we can see from the data, there is no obvious way to say which session any given event belongs to. Before we can perform the session analysis, we need to determine which session a given event belongs to.
This we can achieve using a little bit of ETL on data.
ETL in Tinace
ETL is part of any data analysis pipeline. For the above data, we cannot ask the data team to update their schema to include session indicator as it’s specific to our analysis. Unlike many BI tools out there, Tinace has a full fledged ETL built in to the tool, which allows the user to perform use case specific ETL on the data without touching any of the source data.
Tinace supports point and click ETL for simple transformations
Tinace also supports advanced transformations using SQL and Python
In this post, we will be discussing how to use SQL of Tinace to figure out the sessions in our event data.
Figuring out when Session Starts
First we need to figure out when a session starts. We can figure this out using the LAG function of SQL to calculate the same.
select * , LAG(event_at,1) OVER
(PARTITION BY user_id order by event_at) AS
last_event
from data
The above code figures out the last event time and adds a new column to the dataset. This column will be only added to Tinace data, and not to the original table. By not touching the original table, Tinace users are free to any changes to the data without worrying about the integrity of the source data and effecting the downstream systems.
Now the data will look as below
Now we have figured out when the session starts and added it as a new column. As the next step we need to add additional code to figure out the unique sessions. We will discuss it in the next post.
