big-data-fc / project Goto Github PK

• The work rate is an integer value expressed in [attacking_work_rate, defense_work_rate], with values ranging [0,100]. It should be rescaled as needed
• convert work rate scale of post-15 datasets (from 0 to 2) to pre-15 work rate scale (from 0 to 100)

Decide how to treat football teams inconsistencies; the most evident ones are:

• AC Milan ==> Milan
• Hellas Verona ==> Verona
• U.S. Sassuolo Calcio ==> Sassuolo

For example:

• AC Milan ==> Milan
• FC Barcelona ==> Barcelona
• \1. FC Köln ==> Köln
• SV Darmstadt 98 ==> SV Darmstadt

Things to remove:

• AC,AS,US,UD,FC
• 1. FC,SC,spVgg,DSC,MSV,VfL,SV,1. FSV,RB,TSV,TSG
• AD,BP,CF,CA,C.D.,CD,CE,RCD,RC,SD,UE

When run in colab, the notebook should automatically:

• download and install pyspark
• download the needed data from google drive (with gdown)
• set the appropriate paths

This way, we'd have much more data points (more than 50k)

• Delete unused variables
• Do garbage collection when needed

It breaks the colab environment. Can be safely removed, since interactive plots are not needed anymore.

Taking into account substitute players may negatively affect teams features, considering that, most often than not, "bench" players ain't that good, w.r.t. starting XI players