def __init__(self,):
self.num_dancers = 0 # number of connected dancers
self.is_idle = True
self.counter = 0
self.init_counter = 24
self.start_time = time.time()
dance_model_path = "model_weights.json"
dance_scaler_path = "dnn_std_scaler.bin"
self.ml = ML(
dance_scaler_path=dance_scaler_path, dance_model_path=dance_model_path,
)
self.skip_initial_readings_cnt = [[50, 50], [50, 50], [50, 50]]
def test_curated_playlist(self):
os.mkdir("1")
tracks_to_classify = [{
'url':
"https://p.scdn.co/mp3-preview/b326e03624cb098d8387e17aa46669edac0d025a?cid=774b29d4f13844c495f206cafdad9c86",
"id": "2takcwOaAZWiXQijPHIx7B"
}]
tracks_ids = ModelHandler().curated_tracks(tracks_to_classify, "1")
assert tracks_ids is None
ML().train_model("test_music", "test_music_2", path_to_save="1/model")
track_ids = ModelHandler().curated_tracks(tracks_to_classify, "1")
assert isinstance(track_ids, list)
shutil.rmtree("1")
def __init__(self):
self.spotify_api = SpotifyAPI()
self.ml = ML()
from intcomm import IntComm
from ML import ML
if __name__ == "__main__":
dance_model_path = "model_weights.json"
dance_scaler_path = "dnn_std_scaler.bin"
ml = ML(
dance_scaler_path=dance_scaler_path,
dance_model_path=dance_model_path,
)
# change this according to your serial port
# 0: "/dev/ttyACM0"
# 1: "/dev/ttyACM1"
# 2: "/dev/ttyACM2"
intcomm = IntComm(0)
while True:
data = intcomm.get_line()
if len(data) == 0 or data[0] != "#":
print("Invalid data:", data)
continue
data = data[1:].split(",")
if len(data) == 10:
yaw, pitch, roll, gyrox, gyroy, gyroz, accx, accy, accz, emg = data
yaw, pitch, roll, gyrox, gyroy, gyroz, accx, accy, accz = (
float(yaw),
float(pitch),
help='Interfaces|LSPs|Nodes|Demands.')
parser.add_argument('property', help='Property.')
parser.add_argument('keys', help='"|" delimited list of keys.')
parser.add_argument('from', help='from date YYMMDD_HHMM_UTC.')
parser.add_argument('to', help='to date YYMMDD_HHMM_UTC.')
parser.add_argument('-keyColumns',
help='list of key columns using comma to separate.')
args = vars(parser.parse_args())
try:
(username, password, server) = parse_url(args['url'])
except (ValueError, TypeError) as e:
print "invalid url"
sys.exit(1)
myml = ML(server, {'username': username, 'password': password})
#r = myml.explore(args['object'], args['filter'], 100, {'SetupBW', 'SourceNode'})
date_pattern = '%y%m%d_%H%M_%Z'
date_from = time.strptime(args['from'], date_pattern)
date_to = time.strptime(args['to'], date_pattern)
if args['keyColumns']:
print myml.time_series(args['object'], args['property'],
args['keys'].split("|"), date_from, date_to,
args['keyColumns'].split(","))
else:
print myml.time_series(args['object'], args['property'],
args['keys'].split("|"), date_from, date_to)
#print myml.time_series("Interfaces", "TraffIn", ["AM_LA_BB2", "TenGigE0/2/2"])
def test_train_model(self):
ml = ML()
ml.train_model("test_music",
"test_music_2",
path_to_save="tester_model")
assert "tester_model" in os.listdir()
def test_classify_tracks(self):
ml = ML()
ml.classify_tracks(['test_music/514q3otlT6HczfChuLDUSa.mp3'],
"tester_model", "test_music")
from flask import Flask, jsonify, Response, request
# from ML import ML
from ML import ML
from DB import DB
from Social import Social
from SessionManager import SessionManager
from flask_cors import CORS
app = Flask(__name__)
cors = CORS(app, resources={r"/*": {"origins": "*"}})
ml = ML()
db = DB()
social = Social([db, ml])
sessMan = SessionManager(db, social)
@app.route('/sessMan/getTweets', methods=['get'])
def runGetTweets():
PID = request.args.get('key')[1:]
count = db.getCountPerDay(PID)[0][0]
print(count)
return sessMan.getTweets(PID, count), 200
@app.route('/sessMan/isRunning', methods=['get'])
def profileIsRunning():
PID = request.args.get('key')[1:]
return str(sessMan.isProfileRunning(PID))
@app.route('/sessMan/stopProfile', methods=['get'])
def edit():
json_list = []
##trump=pd.read_csv("trump_tweets_classified.csv")
tweets = ML()
tweets2 = tweets[[
"id", "Created at", "Screen Name", "Tweet Text", "sentiment", "Subject"
]]
tweets2["Frequency"] = 1
#modifying the date-time string
time_string = tweets2["Created at"].to_list()
time_edited = []
for item in time_string:
item2 = item.replace(" +0000", "")
time_edited.append(item2)
tweets2["Created at"] = time_edited
tweets2['Created at'] = pd.to_datetime(tweets2['Created at'],
infer_datetime_format=True)
tweets2['date'] = tweets2['Created at'].dt.date
#most frequent dates most are from the same day, wil lscreq axis if not all from the same day
date = pd.DataFrame(tweets2.groupby("date")["Frequency"].count())
date = date.sort_values(by='Frequency', ascending=False)
#to keep majority of the graph, only display one day of data
date = date.head(1)
day = date.index.values[0]
edited_time_tweets = tweets2.loc[tweets2["date"] == day, [
"date", "Created at", "Screen Name", "Tweet Text", "sentiment",
"Subject", "Frequency"
]]
edited_time_tweets = edited_time_tweets.dropna()
edited = edited_time_tweets.sort_values(by='Created at')
edited = edited.rename(columns={"Created at": "datetime"})
edited = edited.drop(columns=['date'])
#json positive sentiment
only_positive = edited.loc[edited["sentiment"] == "positive", :]
positive_frequency = only_positive[["datetime", "Frequency", "Subject"]]
positive_frequency_trump = positive_frequency.loc[
positive_frequency["Subject"] == "Trump", :]
positive_frequency_andrews = positive_frequency.loc[
positive_frequency["Subject"] == "Andrews", :]
only_positive_trump = positive_frequency_trump.resample(
'0.05T', on='datetime').sum()
only_positive_trump = only_positive_trump.reset_index()
only_positive_trump["sentiment"] = "positive"
only_positive_andrews = positive_frequency_andrews.resample(
'1H', on='datetime').sum()
only_positive_andrews = only_positive_andrews.reset_index()
only_positive_andrews["sentiment"] = "positive"
#json negative sentiment
only_negative = edited.loc[edited["sentiment"] == "negative", :]
negative_frequency = only_negative[["datetime", "Frequency", "Subject"]]
negative_frequency_trump = negative_frequency.loc[
negative_frequency["Subject"] == "Trump", :]
negative_frequency_andrews = negative_frequency.loc[
negative_frequency["Subject"] == "Andrews", :]
only_negative_andrews = negative_frequency_andrews.resample(
'1H', on='datetime').sum()
only_negative_andrews = only_negative_andrews.reset_index()
only_negative_andrews["sentiment"] = "negative"
only_negative_trump = negative_frequency_trump[["datetime", "Frequency"]]
only_negative_trump = negative_frequency_trump.resample(
'0.05T', on='datetime').sum()
only_negative_trump = only_negative_trump.reset_index()
only_negative_trump["sentiment"] = "negative"
frames = [only_positive_trump, only_negative_trump]
sentiment_concat = pd.concat(frames)
sentiment_concat['datetime'] = sentiment_concat['datetime'].astype(str)
sentiment_group = sentiment_concat.groupby(['sentiment', "datetime"])
sentiment_group_2 = pd.DataFrame(sentiment_group["Frequency"].sum())
results1 = defaultdict(lambda: defaultdict(dict))
for index, value in sentiment_group_2.itertuples():
for i, key in enumerate(index):
if i == 0:
nested = results1[key]
elif i == len(index) - 1:
nested[key] = value
else:
nested = nested[key]
frames = [only_positive_andrews, only_negative_andrews]
sentiment_concat = pd.concat(frames)
sentiment_concat['datetime'] = sentiment_concat['datetime'].astype(str)
sentiment_group = sentiment_concat.groupby(['sentiment', "datetime"])
sentiment_group_2 = pd.DataFrame(sentiment_group["Frequency"].sum())
results2 = defaultdict(lambda: defaultdict(dict))
for index, value in sentiment_group_2.itertuples():
for i, key in enumerate(index):
if i == 0:
nested = results2[key]
elif i == len(index) - 1:
nested[key] = value
else:
nested = nested[key]
tweets = {"Trump": results1, "Andrews": results2}
json_list.append(tweets)
trump2 = tweets2.loc[tweets2["Subject"] == "Trump", :]
trump_sentiment = trump2.groupby(['sentiment'])["Frequency"].count()
total_sentiment = trump2["sentiment"].count()
percentage_sentiment_trump = ((trump_sentiment / total_sentiment) *
100).round()
percentage_trump = percentage_sentiment_trump.to_dict()
andrews2 = tweets2.loc[tweets2["Subject"] == "Andrews", :]
andrews_sentiment = andrews2.groupby(['sentiment'])["Frequency"].count()
total_sentiment = andrews2["sentiment"].count()
percentage_sentiment_andrews = ((andrews_sentiment / total_sentiment) *
100).round()
percentage_andrews = percentage_sentiment_andrews.to_dict()
percentage_dict = {
"Trump": percentage_trump,
"Andrews": percentage_andrews
}
json_list.append(percentage_dict)
example = trump2.head(5)
screen_name = example["Screen Name"].to_list()
tweet_text = example["Tweet Text"].to_list()
res = {screen_name[i]: tweet_text[i] for i in range(len(screen_name))}
example2 = andrews2.head(5)
screen_name2 = example2["Screen Name"].to_list()
tweet_text2 = example2["Tweet Text"].to_list()
res2 = {screen_name2[i]: tweet_text2[i] for i in range(len(screen_name))}
tweet_dict = {"Trump": res, "Andrews": res2}
json_list.append(tweet_dict)
return json_list
from ML import ML
'''
data found here: http://files.pushshift.io/reddit/
API docs here: https://github.com/pushshift/api
don't request more than ~1/sec. Each request can return <500 terms.
getting 1000 posts each for 100 users each from 4 subreddits (i.e., 800 requests)
took 48 minutes, so ~4s/request. Hmmm.
'''
ml = ML('/home/declan/Documents/code/reddit_DS/savedat/nyc_losangeles_unitedkingdom_greece_indonesia_japan_newzealand_1000users_24bins_00-16-17')
#ml = ML('/home/declan/Documents/code/reddit_DS/savedat/rizki_dat1/')
ml.addExtraDataSets('/home/declan/Documents/code/reddit_DS/savedat/rizki_dat1/')
#ml.addExtraDataSets('/home/declan/Documents/code/reddit_DS/savedat/rizki_dat2/')
#ml.prettyPrintDB(ml.df)
#ml.postAvgTimesByRegion()
#ml.simpleLinReg()
#ml.cyclicMetricSGD(alpha=10**-4, timesteps=4000, show_plot=False)
ml.NN1()
exit(0)
ml.cyclicMetricSGD(alpha=10**-3, timesteps=4000, show_plot=False)
ml.cyclicMetricSGD(alpha=10**-3, timesteps=40000, show_plot=False)
ml.cyclicMetricSGD(alpha=10**-5, timesteps=40000, show_plot=False)
