Last active
October 14, 2019 18:53
-
-
Save oleg-agapov/803f035aeeaeeee26eb74401b31dec7d to your computer and use it in GitHub Desktop.
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| # df_raw - input dataset | |
| def parse_and_clean(data_frame: pd.DataFrame) -> pd.DataFrame: | |
| # parse json | |
| df = data_frame.join(data_frame["user_json"].apply(json.loads).apply(pd.Series)) | |
| df["user_json"] = df["user_json"].apply(lambda x: x.replace('\n','')) | |
| # explode visits | |
| df2 = pd.DataFrame({ | |
| "uid": df.uid.repeat(df.visits.str.len()), | |
| "sites" : np.concatenate(df.visits.values)} | |
| ).reset_index() | |
| # split columns | |
| df3 = pd.DataFrame([md for md in df2.sites]) | |
| df3["uid"] = df2.uid | |
| df3["url"] = df3["url"].apply(lambda x: x.replace("\n", "")).apply(lambda x: x.replace("\r", "")) | |
| # join initial DF | |
| df = pd.merge(df3, df, how="left", on="uid")[["gender", "age", "uid", "url", "timestamp"]] | |
| df["domain"] = df.url.apply(url_to_domain) | |
| df = df.dropna() | |
| # %timeit pd.to_datetime(df.timestamp.head(), unit="ms") | |
| df.timestamp = pd.to_datetime(df.timestamp, unit="ms") | |
| return df | |
| def url_to_domain(url: str) -> str: | |
| url = re.sub('(http(s)*://)+', 'http://', url) | |
| parsed_url = urlparse(unquote(url.strip())) | |
| if parsed_url.scheme not in ['http','https']: return None | |
| netloc = re.search("(?:www\.)?(.*)", parsed_url.netloc).group(1) | |
| if netloc is not None: return str(netloc).strip().encode('utf-8').decode('utf-8') | |
| return None | |
| def get_domains(input_json): | |
| visits = json.loads(input_json)["visits"] | |
| domains = [] | |
| for visit in visits: | |
| domains.append(url_to_domain(visit["url"])) | |
| return domains | |
| def get_visits_by_hour(input_json): | |
| visits = json.loads(input_json)["visits"] | |
| visits_vector = [] | |
| for h in range(24): | |
| visits_vector.append(0) | |
| for visit in visits: | |
| visit_hour = datetime.datetime.fromtimestamp(int(visit["timestamp"])/1000).hour | |
| visits_vector[visit_hour] += 1 | |
| return visits_vector | |
| train = df_raw.iloc[:100].copy() | |
| train["hour"] = train.user_json.apply(get_visits_by_hour) | |
| train["domains"] = train.user_json.apply(get_domains).apply(' '.join) |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment