top of page
Home: Welcome

DATA ENTHUSIAST

"Maybe stories are just data with a soul."

I am Sanjit Sisodiya - A Data Enthusiast. I am passionate about deriving insights from data and can make stories speak through visualization and dashboards. Over the past four years of my work experience, I have developed a belief in the process of working with data > extracting information >  taking action. Here, is the blog of some of my work involving machine learning and predictive analytics.

TECHNICAL SKILLS:

  • SQL

  • R

  • Python

  • Tableau

  • SPSS

  • Google Analytics

 

Search

ADVANCED NBA STATS ANALYSIS

  • Writer: Sanjit_3282
    Sanjit_3282
  • Nov 1, 2018
  • 2 min read

Updated: Feb 19, 2019

At the start of the 2013-14 season, the NBA implemented player tracking systems into team arenas to track player movements. This provided a wealth of smarter data like passes, secondary assists, dribbles per touch etc. to give more insights into different style of team’s play. These stats can show which teams play a more unselfish style of play with ball and player movement (Golden State Warriors) compared to other teams who have ball dominators and are more isolation based (Houston Rockets).



NBA: James Harden's dunk vs Warriors

We pulled these data from nba.com, adjusted these data per 100 possessions and explored their associations with team performance. The hypothesis was that more ball movement and unselfish play will lead to higher win %, whereas isolation and less ball movement will reduce team’s winning performance and our target variables being Offensive Rating and Win Percentage.


Using linear regression in R, we identified statistically significant variables in predicting offensive rating and teams were clustered into groups based on their style of play using the variables highlighted from regression. Using SPSS modeler, we determined 3 as the optimal number of clusters for our data, with clusters being Ball-Movement, Moderate and Isolation based on their style of play.


Our research showed that Adjusted assists and Potential assists were most significant in predicting offensive rating, whereas average dribbles per touch was a better description of isolation teams. ANOVA test showed there is not a statistically significant difference for offensive rating between clusters, but there is for win %. Teams with higher ball movement have a higher win % on average, but that’s not always the case as shown by Houston Rockets with James Harden and Chris Paul. To conclude, The NBA is still a player driven league, and identifying the effect of specific players in this regard is an important next step.


Below is the video of report for your reference. This was a team project and my team members were Evan DeCastros, Junyi Huang and Yichen Pan.







 
 
 

Comments

Home: Blog2
Home: GetSubscribers_Widget

©2018 by Sanjit Sisodiya. Proudly created with Wix.com

bottom of page