Reference no: EM132271207

Assignment - KOBE BRYANT SHOT SELECTION

OVERVIEW: Kobe Bryant marked his retirement from basketball by scoring 60 points in his final game as a member of the Los Angeles Laker team on Wednesday, April 12, 2016. Starting to play professional basketball at the age of 17, Kobe earned the sport's highest accolades throughout his long career. Using 20 years of data on Kobe's shots made and shots missed, can you predict which shots will be successful?

DATA: The original data set contains the location and circumstances of every shot attempted by Bryant during his 20-year career. Your task is to predict whether the basket went in (shot_made_flag = 1) or missed (shot_made_flag = 0). The data for estimation is in Kobe.xlsx.

For this exercise, 5000 of the shot_made_flags have been removed from the original data set and are shown as missing values in the project2Pred.xlsx file. These are the test set shots for which you must submit a classification. You are provided a sample classification file, project2Pred.xlsx with the shot_ids needed for your predicted classification. Provide you predicted classifications in this file and submit both your paper and the prediction file. I have the actual values of the shot_made_flag for these missing shot_ids and will evaluate the classifications. Your goal is to provide the best predictions possible.

Each group is on the honor system to not use any information outside of the dataset to predict each of the missing shot flags.

DATA CONTINUED

The field names are given below (Data descriptions are available in Kaggle):

action_type combined_shot_type game_event_id game_id lat - court location identifier (latitude) loc_x - court location identifier (x/y axis) loc_y- court location identifier (x / y axis) lon - court location identifier (longitude) minutes_remaining - (in period) period playoffs season  seconds_remaining attendance avgnoisedb - avg noise in arena (decibels)

shot_distance shot_made_flag (this is what you are predicting) shot_type shot_zone_area shot_zone_basic shot_zone_range team_id team_name game_date matchup opponent shot_id arena_temp (oF)

DELIVERABLE: Submit a paper with an 8 page limit with a separate Appendix up to 5 pages. Code should be in a second appendix and can be as long as necessary. A separate file with predicted classifications also should be submitted.

PAPER REQUIREMENTS -

Introduction

Data Description

Exploratory Data Analysis

• Address the need for any potential transformations.
• Address and identify outliers.
• Address and identify any multicollinearity.

Build models to provide arguments and evidence for or against the propositions below:

• The odds of Kobe making a shot decrease with respect to the distance he is from the hoop. If there is evidence of this, quantify this relationship. (CIs, plots, etc.).
• The probability of Kobe making a shot decreases linearly with respect to the distance he is from the hoop. If there is evidence of this, quantify this relationship. (CIs, plots, etc.).
• The relationship between the distance Kobe is from the basket and the odds of him making the shot is different if they are in the playoffs. Quantify your findings with.

Build a predictive model to classify shots as missed or made. You should produce at least 1 of each type of model:

• A logistic regression model.
• A Linear Discriminant Analysis (LDA) model.

Evaluation: Compare each competing models with the AUC, Mis-Classification Rate, Sensitivity, Specificity and objective / loss function. The log loss function of the model should be used to assess the model fit:

-1/N _i=1∑^N[y_ilog p_i + (1 - y_i)log(1 - p_i)].

Where N is the total number classifications, yi is the shot_made_flag and pi is the probability from the model of each outcome (shot made or shot missed.)

Note - Need A SAS programming assignment done. All relevant info in the zip files.

Attachment:- Kobe-data file.rar

Attachment:- Assignment Files.rar