def getCircuitRaceList(filepath):
"""
Parameters
----------
filepath : string
The filepath relative to the parentPath
Returns
-------
dict{key:[list]}
{circuitID: [raceID]}
"""
circuit_Data = getDataset(filepath)
dataObtained = circuit_Data.next()
RecentTracksList = getRecentTracks()
print "Data from file: " + str(dataObtained)
print "Data Extracted: " + 'circuitID ' + " raceID"
circuitDict_raceList = defaultdict(
lambda: []) # initialise the dictionary value to an empty list
for race_circuit in circuit_Data:
# print race_circuit
if (race_circuit[2] in RecentTracksList):
circuitDict_raceList[race_circuit[2]].append(
race_circuit[0]) # {circuitID: [raceID]}
return circuitDict_raceList
def getPitStops(filepath):
"""
Parameters
----------
filepath: string
The directory to the file relative to the parentPath
# [raceId, driverId, stop, lap, time, duration, milliseconds]
Returns
-------
tuple
(headings, data, dictionary)
"""
RecentDriversList = getRecentDrivers(2000)
circuitRaceList = getCircuitRaceList("races.csv")
circuitRaceKeys = circuitRaceList.keys()
PitStopsData = getDataset(filepath)
dataObtained = PitStopsData.next()
Data = []
for data in PitStopsData:
if (ifExist(data[1], RecentDriversList)):
for key in circuitRaceKeys:
if (data[0] in circuitRaceList[key]):
Data.append(data)
# Data = list(PitStopsData)
print "Data from file: " + str(dataObtained)
# print "Data Extracted: " + 'Year ' + " raceID"
_ret_dict = defaultdict(lambda: defaultdict(lambda: 0.0))
for d in Data:
# {raceId, driverId :{ stop: pit time }}
_ret_dict[str(d[0]) + "," + str(d[1])][d[2]] = d[6]
return (dataObtained, Data, _ret_dict)
def getPreprocessedData(filepath):
"""
Parameters
----------
filepath: string
The directory to the file relative to the parentPath
# [raceId, driverId, constructorId, rank, points, grid, position, positionOrder, positionText , firstLapChange, pitStops]
Returns
-------
tuple
(headings, data, dictionary)
"""
RecentDriversList = getRecentDrivers(2000)
circuitRaceList = getCircuitRaceList("races.csv")
circuitRaceKeys = circuitRaceList.keys()
PreprocessedData = getDataset(filepath)
dataObtained = PreprocessedData.next()
# Data = list(PreprocessedData)
Data = []
for data in PreprocessedData:
if (ifExist(data[1], RecentDriversList)):
for key in circuitRaceKeys:
if (data[0] in circuitRaceList[key]):
Data.append(data)
print "Data from file: " + str(dataObtained)
return (dataObtained, Data)
def getDataFromResults(filepath):
"""
Parameters
----------
filepath: string
The directory to the file relative to the parentPath
# [raceID, driverID, constructorID, rank, points, grid, postion, positionOrder]
Returns
-------
tuple
(headings, data)
"""
RecentDriversList = getRecentDrivers(2000)
circuitRaceList = getCircuitRaceList("races.csv")
circuitRaceKeys = circuitRaceList.keys()
ResultData = getDataset(filepath)
dataObtained = ResultData.next()
Data = []
# Data = list(ResultData)
for data in ResultData:
if (ifExist(data[1], RecentDriversList)):
for key in circuitRaceKeys:
if (data[0] in circuitRaceList[key]):
Data.append(data)
print "Data from file: " + str(dataObtained)
# print "Data Extracted: " + 'Year ' + " raceID"
# yearDict_raceList = defaultdict(lambda:[]) # initialise the dictionary value to an empty list
# dataObtained.append("firstLapChange")
# for d in Data:
# d.append(int(d[5]) - int(d[7]))
return (dataObtained, Data)
def getCircuitInfoList(filepath):
"""
Parameters
----------
filepath : string
The filepath relative to the parentPath
Returns
-------
dict{key:[list]}
{circuitID: [track information]}
"""
circuit_Data = getDataset(filepath)
dataObtained = circuit_Data.next()
print "Data from file: " + str(dataObtained)
print "Data Extracted: " + 'circuitID ' + " raceInfo"
circuitDict_InfoList = defaultdict(lambda: np.array([
])) # initialise the dictionary value to an empty list
for circuitInfo in circuit_Data:
# print race_circuit
for i in xrange(len(circuitInfo)):
if ((i > 8) & (i < 1009)):
# print circuitInfo[i]
# if( circuitInfo[i] != ""):
circuitDict_InfoList[circuitInfo[1]] = np.append(
circuitDict_InfoList[circuitInfo[1]],
np.int(circuitInfo[i]))
# else:
# print circuitInfo[1]
return circuitDict_InfoList
def getRaceDriverConstRankPts(filepath): """ Parameters ---------- filepath : string The filepath relative to the parentPath Returns ------- [raceID, driverID, constructorID, rank, points] """ cons_Data = getDataset(filepath) dataObtained = cons_Data.next() print "Data Extracted: " + str(dataObtained) return list(cons_Data)
def getLapTimes(filepath):
"""
Parameters
----------
filepath: string
The directory to the file relative to the parentPath
# [raceId, driverId, lap, position, time, milliseconds]
Returns
-------
tuple
(headings, data, dictionary, dictionary of dictionary)
"""
RecentDriversList = getRecentDrivers(2000)
circuitRaceList = getCircuitRaceList("races.csv")
circuitRaceKeys = circuitRaceList.keys()
LapTimesData = getDataset(filepath)
dataObtained = LapTimesData.next()
# Data = list(LapTimesData)
Data = []
for data in LapTimesData:
if (ifExist(data[1], RecentDriversList)):
for key in circuitRaceKeys:
if (data[0] in circuitRaceList[key]):
Data.append(data)
print "Data from file: " + str(dataObtained)
# {raceID: [[driverID, lap#, laptime]]}
_ret_dict = defaultdict(lambda: [])
# {raceID: {lap#: [driverID, laptime]}}
_ret_dict_dict = defaultdict(lambda: defaultdict(lambda: []))
for d in Data:
# {raceId, driverId :{ lap: laptime]]
# _ret_dict[str(d[0]) +"," + str(d[1])][d[2]] = d[5]
_ret_dict[int(d[0])].append(np.array([int(d[1]),
int(d[2]),
int(d[5])]))
# {raceID: {lap#: [driverID, laptime]}}
_ret_dict_dict[int(d[0])][int(d[2])].append([int(d[1]), int(d[5])])
# keys = sorted(_ret_dict_dict.keys(), lambda _key: int(_key))
keys = _ret_dict_dict.keys()
for key in keys:
_keys = _ret_dict_dict[key].keys()
for _key in _keys:
_ret_dict_dict[key][_key].sort(key=itemgetter(1))
# print _ret_dict_dict[key][_key]
return (dataObtained, Data, _ret_dict, _ret_dict_dict)
def getDriverIDName(filepath):
"""
filepath: drivers.csv
Returns:
--------
dictionary:
{DriverID: DriverName}
"""
driversData = getDataset(filepath)
dataObtained = driversData.next()
print "Data from file: " + str(dataObtained)
print "Data Extracted: " + 'driverID ' + " driverName"
driverIDNameDict = defaultdict(
lambda: []) # initialise the dictionary value to an empty list
for data in driversData:
driverIDNameDict[data[0]] = data[1] + " " + data[2]
return driverIDNameDict
def getDriverIDName(filepath):
"""
Parameters
----------
filepath : string
The filepath relative to the parentPath
Returns
-------
dict{key:string}
{ConstructorID: ConstructorName}
"""
driver_Data = getDataset(filepath)
# print list(cons_Data)
dataObtained = driver_Data.next()
print "Data Extracted: " + str(dataObtained)
ret_dict = {}
for _driver in driver_Data:
ret_dict[_driver[0]] = _driver[1] + " " + _driver[2]
return ret_dict
def getYearRaceID(filepath):
"""
Parameters
----------
filepath: string
The directory to the file relative to the parentPath
Returns
-------
dictionary of list
{year: [raceID]}
"""
YearRace_Data = getDataset(filepath)
dataObtained = YearRace_Data.next()
print "Data from file: " + str(dataObtained)
print "Data Extracted: " + 'Year ' + " raceID"
yearDict_raceList = defaultdict(lambda:[]) # initialise the dictionary value to an empty list
for race_year in YearRace_Data:
# print race_circuit
yearDict_raceList[race_year[1]].append(race_year[0]) # {circuitID: [raceID]}
return yearDict_raceList
def getConstructorRaceList(filepath):
"""
Parameters
----------
filepath : string
The filepath relative to the parentPath
Returns
-------
dict{key:[list]}
{constructorID: [raceID]}
"""
cons_Data = getDataset(filepath)
dataObtained = cons_Data.next()
print "Data Extracted: " + str(dataObtained)
consDict_raceList = defaultdict(
lambda: []) # initialise the dictionary value to an empty list
for race_cons in cons_Data:
consDict_raceList[race_cons[1]].append(
race_cons[0]) # {constructorID: [raceID]}
return consDict_raceList
def getDataFromTracks(filepath):
"""
Parameters
----------
filepath: string
The directory to the file relative to the parentPath
# [raceID, driverID, constructorID, rank, points, grid, postion, positionOrder]
Returns
-------
tuple
(headings, data)
"""
TrackData = getDataset(filepath)
# TrackData.next()
dataObtained = TrackData.next()
Data = list(TrackData)
print "Data from file: " + str(dataObtained)
# print "Data Extracted: " + 'Year ' + " raceID"
# yearDict_raceList = defaultdict(lambda:[]) # initialise the dictionary value to an empty list
# dataObtained.append("firstLapChange")
# for d in Data:
# d.append(int(d[5]) - int(d[7]))
return (dataObtained, Data)
