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 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 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 filterDataAccordingToRecentDrivers2(dataset): _ret_dataset = [] RecentDriverList = getRecentDrivers(2000) _temp_headings = dataset[0] _ret_dataset.append(_temp_headings) for data in dataset: if(data[1] in RecentDriverList): _ret_dataset.append(data) return _ret_dataset
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)