def eval(predictions, type='TWEET', predictToFile='predictionsTmp.json'):
out_file = open(predictToFile, "w")
for i in range(predictions.shape[0]):
id = testIDs[i]
prediction = predictions[i]
bestPlace = np.argmax(prediction)
placeName = colnames[bestPlace]
my_dict = {
'hashed_tweet_id': id,
'city': placeName,
'lat': placeMedian[placeName][0], # 20.76
'lon': placeMedian[placeName][1] # 69.07
}
json.dump(my_dict, out_file)
out_file.write("\n")
out_file.close()
evaluate_submission(predictToFile, goldFile, type)
def evalMax(predictions, type='USER', predictToFile='predictionsUserTmp.json'):
out_file = open(predictToFile, "w")
for userHash in set(testUserIds):
indices = [i for i, x in enumerate(testUserIds) if x == userHash]
prediction = predictions[indices] #Get all predictions for that user
bestPlace = np.argmax(prediction) % prediction.shape[1]
placeName = colnames[bestPlace]
my_dict = {
'hashed_user_id': userHash,
'city': placeName,
'lat': placeMedian[placeName][0], # 20.76
'lon': placeMedian[placeName][1] # 69.07
}
# print(placeName +" " +instance.text)
json.dump(my_dict, out_file)
out_file.write("\n")
out_file.close()
evaluate_submission(predictToFile, goldFile, type)
out_file = open("predictions.json", "w")
for id in testIDs:
my_dict = {
'hashed_tweet_id': id,
'city': 'jakarta-04-id',
'lat': lat, # 20.76
'lon': lon # 69.07
}
# print(placeName +" " +instance.text)
json.dump(my_dict, out_file)
out_file.write("\n")
out_file.close()
from geoEval import evaluate_submission
evaluate_submission('predictions.json', 'test/test_labels/oracle.tweet.json',
'TWEET')
###Evaluate simple baseline for User predictions
f = open(testUser)
testUsernames = set()
for line in f:
instance = parseJsonLine(line)
testUsernames.add(instance.userName)
out_file = open('predictionsUser.json', "w")
for userHash in testUsernames:
my_dict = {
'hashed_user_id': userHash,
'city': 'jakarta-04-id',
'lat': lat,