neuro-scouting-mlb

-Author: David Abrahams, 11/5/2014

• How to Run - Using a Python interpreter run executable.py from the src directory. Place your data files in a separate directory adjacent to src.

• Python package dependencies - numpy matplotlib

• How the script works -

1. The user enters the directory where the inning.xml files are located.

2. The script parses through each file ending in 'inning.xml'

3. The script stores every 'at_bat' element in a list

4. The script builds a dictionary that maps from pitch-sequence to a list of result counts by doing the following: 4a. Extract the final three pitches of an at bat. 4b. Check if the dictionary has a key for that specific three pitch series. 4c. If it does, increment the correct result count. Example (at bat strikeout ending in FA, FA, CU):

   sequence_results_dict[(FA, FA, CU)] = {'Hit': 14, 'Hit (out)': 29, 'Strikeout': 10, 'Walk': 3, 'Other': 1} {'Hit': 14, 'Hit (out)': 29, 'Strikeout': 10, 'Walk': 3, 'Other': 1} ==> {'Hit': 14, 'Hit (out)': 29, 'Strikeout': 11, 'Walk': 3, 'Other': 1}

4d. If it doesn't, create a new key-value pair in the dictionary and increment the correct value. Example:

sequence_results_dict[(FA, FA, CU)] = {'Hit': 0, 'Hit (out)': 0, 'Strikeout': 1, 'Walk': 0, 'Other': 0}

5. The user enters a three pitch sequence, and using that sequence as a key, the script finds the result counts for that sequence. Example:

   sequence_results_dict[(FA, FA, CU)] = {'Hit': 14, 'Hit (out)': 29, 'Strikeout': 11, 'Walk': 3, 'Other': 1}

6. The script displays the result counts in a bar graph.

• Code organization - executable.py: contains the main method for the program, calls interact_with_user() in the user_interactions.py module user_interactions.py: the program’s ‘Controller’, contains all user interactions and inputs data_visualizer.py: the program’s ‘View’, displays pitch results to the user using the matplotlib library data_reader.py: loads data from a directory into the program data_processor.py: the program’s ‘Model’, takes the data loaded from data_reader.py and manipulates it into a dictionary that maps from an end-of-at-bat pitch sequence to result counts of that sequence.

• Optimizations - Rather than taking in a three-pitch user input and then parsing through the all the data looking for at-bats ending in those three pitches, the program parses through all the data once at the beginning and creates a dictionary that maps from three pitch sequences to result counts. That way, when the user inputs three pitches, the program simply gets the value associated with the three pitch key in the stored dictionary.