pysoccer.algorithms.playerank.features package

Submodules

pysoccer.algorithms.playerank.features.abstract module

class pysoccer.algorithms.playerank.features.abstract.Aggregation

Bases: object

Defines the methods to aggregate one/more collection of features for each match it have to provide results as a dataframe, e.g. it is used to compute relative feature for each match match -> team (or entity) -> featureTeam - featureOppositor

abstract aggregate()

Merge the collections of feature and aggregate by match and team, computing the relative value for each team e.g. match -> team (or entity) -> featureTeam - featureOppositor

Returns

a dataframe

abstract property get_features

abstract property set_features

Set the list of collection to use for relative features computing

Parameters

collection_list – list of collection. E.g. we could have a collection of quality features, one for quantity features, one for goals scored etc

class pysoccer.algorithms.playerank.features.abstract.Feature

Bases: object

Class to wrap all the scripts/method to aggregate features from the database

abstract createFeature(collectionName, param)

Method to define how a feature/set of features is computed.

Parameters

• param – contains eventual parameters for querying database competion, subset of teams, whatever

• collectionName – best practice: features have to be stored into a collection of documents in the form: {_id: {match: (numeric) unique identifier of the match, name: (string) name of the feature, entity: (string) name of the entity target of the aggregation. It could be teamId, playerID, teamID + role or whatever significant for an aggregation}, value: (numeric) the count for the feature}

Returns

the name of the collection where the features have been stored

pysoccer.algorithms.playerank.features.centerOfPerformanceFeature module

class pysoccer.algorithms.playerank.features.centerOfPerformanceFeature.centerOfPerformanceFeature

Bases: pysoccer.algorithms.playerank.features.abstract.Feature

createFeature(serialized_events, players_file, select=None)

Compute centerOfPerformanceFeatures

Parameters

• serialiazed_events – folder path of events file

• players_file – file path of players data file

• select – function for filtering matches collection. Default: aggregate over all matches

• entity – it could either ‘team’ or ‘player’. It selects the aggregation for qualityFeatures among teams or players qualityfeatures. Note: aggregation by team is exploited during learning phase, for features weights estimation, while aggregation by players is involved for rating phase.

Returns

list of json docs dictionaries in the format: {matchId : int , entity : int, feature: string , value}

pysoccer.algorithms.playerank.features.goalScoredFeatures module

class pysoccer.algorithms.playerank.features.goalScoredFeatures.goalScoredFeatures

Bases: pysoccer.algorithms.playerank.features.abstract.Feature

Goals scored by each team in each match

createFeature(serialized_matches, select=None)

Stores qualityFeatures on database

Parameters

• serialized_matches – file path of matches file

• select – function for filtering matches collection. Default: aggregate over all matches

Returns

list of documents in the format: match: matchId, entity: team, feature: feature, value: value

pysoccer.algorithms.playerank.features.matchPlayedFeatures module

class pysoccer.algorithms.playerank.features.matchPlayedFeatures.matchPlayedFeatures

Bases: pysoccer.algorithms.playerank.features.abstract.Feature

createFeature(serialized_matches, players_file, select=None)

It computes, for each player and match, total time (in minutes) played, goals scored and

Parameters

• serialized_matches – folder with json files corresponding to matches data

• select – function for filtering matches collection. Default: aggregate over all matches

Returns

a collection of documents in the format _id-> {‘match’: this.wyId, ‘player’ : player, ‘name’: ‘minutesPlayed’|’team’|’goalScored’|’timestamp’},value: <float>|<string>;

pysoccer.algorithms.playerank.features.plainAggregation module

class pysoccer.algorithms.playerank.features.plainAggregation.plainAggregation

Bases: pysoccer.algorithms.playerank.features.abstract.Aggregation

Merge features for each player and return a data frame match -> team (or entity) -> feature (playerank, timestamp, team, etc..)

aggregate(to_dataframe=False)

Merge the collections of feature and aggregate by match and team, computing the relative value for each team e.g. match -> team (or entity) -> featureTeam - featureOppositor

Returns

a dataframe

get_aggregated_collection()

get_features()

set_aggregated_collection(collection)

set_features(collection_list)

Set the list of collection to use for relative features computing

Parameters

collection_list – list of collection. E.g. we could have a collection of quality features, one for quantity features, one for goals scored etc

pysoccer.algorithms.playerank.features.playerankFeatures module

class pysoccer.algorithms.playerank.features.playerankFeatures.playerankFeatures

Bases: pysoccer.algorithms.playerank.features.abstract.Feature

Given a method to aggregate features and the corresponding weight of each feature, it computes playerank for each player and match

:param features weights, computed within learning phase of playerank framework

Returns

a collection of json documents in the format: {match:match_id, name: ‘playerankScore’, player:player_id, value: playerankScore(float)}

createFeature(weights_file)

Method to define how a feature/set of features is computed.

Parameters

• param – contains eventual parameters for querying database competion, subset of teams, whatever

• collectionName – best practice: features have to be stored into a collection of documents in the form: {_id: {match: (numeric) unique identifier of the match, name: (string) name of the feature, entity: (string) name of the entity target of the aggregation. It could be teamId, playerID, teamID + role or whatever significant for an aggregation}, value: (numeric) the count for the feature}

Returns

the name of the collection where the features have been stored

get_features()

set_features(collection_list)

pysoccer.algorithms.playerank.features.qualityFeatures module

class pysoccer.algorithms.playerank.features.qualityFeatures.qualityFeatures

Bases: pysoccer.algorithms.playerank.features.abstract.Feature

Quality features are the count of events with outcomes. E.g. - number of accurate passes - number of wrong passes …

createFeature(serialized_events, players_file, entity='team', select=None)

Compute qualityFeatures

Parameters

• serialized_events – file path of events file

• select – function for filtering events collection. Default: aggregate over all events

• entity – it could either ‘team’ or ‘player’. It selects the aggregation for qualityFeatures among teams or players qualityfeatures

Returns

list of dictionaries in the format: matchId -> entity -> feature -> value

pysoccer.algorithms.playerank.features.relativeAggregation module

class pysoccer.algorithms.playerank.features.relativeAggregation.relativeAggregation

Bases: pysoccer.algorithms.playerank.features.abstract.Aggregation

Compute relative feature for each match match -> team (or entity) -> featureTeam - featureOpponents

aggregate(to_dataframe=False)

Compute relative aggregation: give a set of features it compute the A-Bvalue for each entity in each team. Ex: passes for team A in match 111 : 500 passes for team B in match 111 : 300 lead to output: {‘passes’: 200} This method is involved for feature weight estimation phase of playerank framework.

:param to_dataframe : return a dataframe instead of a list of documents

get_features()

set_features(collection_list)

Set the list of collection to use for relative features computing

Parameters

collection_list – list of collection. E.g. we could have a collection of quality features, one for quantity features, one for goals scored etc

pysoccer.algorithms.playerank.features.roleFeatures module

class pysoccer.algorithms.playerank.features.roleFeatures.roleFeatures

Bases: pysoccer.algorithms.playerank.features.abstract.Feature

createFeature(matrix_role_file)

Given the matrix for roles, it computes, for each player and match, the role of a player. A role matrix is a data structure where, given x and y (between 0 and 100), it contains the correspinding roles for a player having center of performance = x,y Role_matrix is computed within learning phase of playerank framework.

Parameters

• role_matrix – file patch for dictionary in the format x->y->role

• feature_lists – lists of features for each player in each match, describing players’ average position

get_features()

set_features(collection_list)

pysoccer.algorithms.playerank.features.wyscoutEventsDefinition module

Module contents