A/B testing - Keep a check on machine learning models’ performance.Perform customer segmentation and targeting.Analyze website traffic and user funnels.The data science team works with the product managers to analyze the following. The choices available are few and limited.įig: Internal Hierarchy of Partner Product Managers Product Management Metabase provides the following means of visualization of data. One of the ways we use this is to forecast how much data will be generated and computed to predict our GCP cost. This further helps us in visualizing risk which translates to policy pricing. We leverage maps visualization for Travel policies to help us identify hot and cold routes. We leveraged Tableau by presenting the information in the form of stories, with multiple views, filters, and in a wide array of layouts and formats. We use Tableau to create: Tableau Dashboards We use data visualization daily, to review things like - the policies sold over months, the amount of Gross Written Premium over time, and detailed partner-product statistics. Tableau has a simple UI and this helps us in creating connections and schedules with minimal effort. We used Tableau Desktop to create data sources. Comparatively, custom SQL with joins performed better since that computation is done on BigQuery. Possibly because of the volume of data in the underlying BigQuery dataset that made it a very heavy task to aggregate and perform a left join. Data Blending: We tried data blending and found it to be quite laggy and unresponsive on Tableau Explorer.These are sent on a daily, weekly, or monthly basis. Subscriptions: Decision-makers subscribe to selected data visualizations.We refresh the data either incrementally or fully. Extract Connections: Extracts are data sources that don’t require live connections, for example, Google Analytics in our use-case.Query execution is handled in the BigQuery. Live Connections: Live connections are set up for tables which have the largest volume of data.We explored the following methods for fetching and manipulating the data: The data is fetched either using a live connection or, using a store that is batch refreshed. Insights from it help us tune and improve our products. Our organization depends on the latest data that is generated. Data processing is handled in Big Query, thus, visuals render faster in Tableau.It is faster for pivoting data and visualization.It supports a large number of custom user functions/calculations.Inbuilt functions enable users to extract data from semi-structured data like JSON fields.Convert dimensions to measures with the click of a button.It is intuitive to use, thus managers and analysts can drag, drop data fields in dimensions and measures.Tableau has enabled our users to use “Investigative Analysis”. We are striving to make this process of data exploration easy for our decision-makers. We realized the core factors needed to drive self-service business intelligence -įig: Core factors for self-service business intelligence Data Exploration and DiscoveryĮxploring and deriving insights from our data help us make quick informed decisions.
#METABASE BIGQUERY UPDATE#
PasarPolis leverages the power of Google BigQuery as a data warehousing platform along with Airflow for data-orchestration. Fig: Photo by Franki Chamaki on Unsplash Current Data Analytics Infrastructure
0 kommentar(er)
