def model_data():
traing_data = s3_bucket_function.s3BucketRead("hmitraningdata",
"hmitraningdata.csv", "csv")
traing_data["Target(Class)"].value_counts()
traing_data["class_name"] = traing_data["Target(Class)"].map({
1:
"Sudden Acceleration",
2:
"Sudden Right Turn",
3:
"Sudden Left Turn",
4:
"Sudden Break",
})
#plt.figure(figsize=(20,10))
#sb.countplot(x = "class_name",data = traing_data)
#plt.show()
X_feature = traing_data.iloc[:, 1:7]
Y_class = traing_data["class_name"]
X_feature.shape
X_train, X_test, Y_train, Y_test = train_test_split(X_feature,
Y_class,
test_size=.30,
random_state=101)
#Total Sample in Training data and Testing data
#print("X_train :{},X_test: {} ,Y_train: {} and Y_test: {}".format(len(X_train),len(X_test),len(Y_train),len(Y_test)))
max_k = int(np.sqrt(len(X_train)))
#max_k
k = list(range(max_k - 1))
len(k)
Ya = Y_test # X_test
list_acc = []
for i in range(1, max_k):
#print(i)
knn_trainer = KNeighborsClassifier(n_neighbors=i)
knn_learner = knn_trainer.fit(X_train, Y_train)
#Prediction for X_test
Yp_knn = knn_learner.predict(X_test)
acc_knn = accuracy_score(Ya, Yp_knn) * 100
#print("acc of knn {} for k {}".format(acc_knn,i))
list_acc.append(acc_knn)
len(list_acc)
list_arr = np.array(list_acc)
max_acc = list_arr.max()
opt_k = list_arr.argmax()
#print(max_acc, opt_k)
#plt.plot(k,list_acc)
#plt.plot(k,list_acc,'r*')
#plt.xlabel("k....>")
#plt.ylabel("acc of k")
#plt.show()
#Selected model
knn_selected = KNeighborsClassifier(n_neighbors=opt_k)
knn_model = knn_selected.fit(X_train, Y_train)
return knn_model
def driver_trip_analysis_function(bucketName, driver_id_maxtimesatmp,
filetype):
real_driver = s3_bucket_function.s3BucketRead("hmimlbucket",
driver_id_maxtimesatmp,
filetype)
Acc = real_driver.iloc[:, 4:7]
Gyro = real_driver.iloc[:, 7:10]
data_acc_gyro = pd.concat([Gyro, Acc], axis=1)
Driver_behaviour = knn_model.predict(data_acc_gyro)
Result_driver = pd.DataFrame({"Prediction": Driver_behaviour})
return Result_driver, real_driver, Driver_behaviour
def real_data():
#Master Data
real_driver_master = s3_bucket_function.s3BucketRead(
"hmimaster", "master", "csv")
real_driver_master = real_driver_master
Acc = real_driver_master.iloc[:, 5:8]
Gyro = real_driver_master.iloc[:, 8:11]
data_acc_gyro = pd.concat([Gyro, Acc], axis=1)
data_acc_gyro = data_acc_gyro
Driver_behaviour = knn_model.predict(data_acc_gyro)
Result_driver = pd.DataFrame({"Prediction": Driver_behaviour})
result_hmi = Result_driver["Prediction"].value_counts()
def calculate_all_raw_file():
file_list = s3_bucket_function.list_of_file_name_from_s3_bucket(
"febappuser")
all_cav_data = {}
for i in range(len(file_list)):
data = s3_bucket_function.s3BucketRead("febappuser", file_list[i],
"txt")
all_cav_data[i] = raw_to_csv(data, file_list[i])
temp = file_list[i]
file_name = re.findall('\D+[\d]+', temp)
file_name = file_name[0] + file_name[1]
s3_bucket_function.s3BucketWrite("febappcsv", file_name,
all_cav_data[i])
return all_cav_data
def merge_result_file(result_file):
result_file = result_file
colums_name = [
'Unnamed: 0', 'sudden_left_turn', 'sudden_acceleration',
'sudden_break', 'sudden_right_turn', 'driver', 'trip_duration',
'speed_max', 'speed_min', 'speed_avg', 'start_lat', 'start_lon',
'end_lat', 'end_lon', 'start_address', 'end_address', 'idle_time',
'final_score'
]
df = pd.DataFrame(columns=colums_name)
for i in range(len(result_file)):
temp = s3_bucket_function.s3BucketRead("hmiresultdriverbehaviourcsv",
result_file[i], "csv")
df = df.append(temp, ignore_index=False)
df = df.drop(["Unnamed: 0"], axis=1)
#temp_value = temp.values.tolist()[0]
#df.insert(i,colums_name,temp_value)
return df