Introduction

“Have you ever seen the view of Los Angeles at four o ‘ clock in the morning?”

— Kobe Bryant .

Background

In this project , we chose Kobe Bryant’s career game data to analyse and try to predict the hit or miss of his shots. It is also a data competition from Kaggle.

We explore all shot attempts of Kobe Bryant throughout his career to predict the hit or miss of some shots, and try to find some underlying information.

Data overview

Above raw data are all from Kaggle

variable	Value_type	Example
action_type	String	Jump Shot,Layup Shot,Driving Dunk Shot…
combined_shot_type	String	Jump Shot,Layup,Dunk,Tip Shot…
game_event_id	Numerical	2,3,4,…
game_id	Numerical	20000012,20000019,20000047…
lat	Numerical	33.2533,33.2713,33.3033…
loc-x	Numerical	-250 – 250
loc-y	Numerical	-50 – 890
lon	Numerical	-118.5198,-118.5178,-118.5158…
minutes_remaining	Numerical	0-11
period	Numerical	1-7
playoffs	Numerical	0,1
season	String	1996-97,1997-98,1998-99…
seconds_remaining	Numerical	0-59
shot_distance	Numerical	0,37,79…
shot_type	String	2PT Field Goal, 3PT Field Goal
shot_zone_area	String	Center(C),Right Side Center(RC),Right Side(R)…
shot_zone_basic	String	Mid-Range,Restricted Area,Above the Break 3…
shot_zone_range	String	Less Than 8 ft.,16-24 ft.,8-16 ft.…
team_id	Numerical	1610612747
team_name	String	Los Angeles Lakers
game_date	Date	2000/10/31,2000/11/1,2000/11/4…
match-up	String	LAL @ ATL,LAL @ BKN,LAL @ BOS…
opponent	String	ATL,BKN,BOS…
shot_id	Numerical	1-30697

Target variable

The target variable is“shot_made_flag”, which means that whether this shot hits. 5,000 of the more than 30,000 pieces of data were selected as test sets, and the label value was hidden. The final job was to predict the possibility of hitting 5,000 shots based on the rest of the data.