Numerai Signals 시작하기: 감정 분석(Sentiment Analysis)

import requests
api_key = "YOUR_API_KEY_HERE"
api_request = f"https://stocknewsapi.com/api/v1?tickers=TSLA&items=1&token={api_key}"
data = requests.get(api_request).json()['data']
print(data)

[
    {
	"news_url": "https://www.barrons.com/articles/fund-bought-tesla-apple-microsoft-stock-sold-att-51613063598",
        "image_url": "https://cdn.snapi.dev/images/v1/a/7/im-298821.jpg",
        "title": "A Huge Fund Bought Tesla, Apple, and Microsoft Stock. Here's What It Sold.",
        "text": "Dutch pension fund PGGM initiated a position in EV-giant Tesla, bought more Apple and Microsoft stock, and sold AT&T stock in the fourth quarter.",
        "source_name": "Barrons",
        "date": "Sun, 14 Feb 2021 07:00:00 -0500",
        "topics": ["paywall"],
        "sentiment": "Neutral",
        "type": "Article",
        "tickers": [
            "AAPL",
            "MSFT",
            "TSLA",
            "T"
        ]
    }

import requests
import pandas as pd

page_cutoff_point = 10
dfs = []
i = 1
while i <= page_cutoff_point:
    api_request = f"https://stocknewsapi.com/api/v1?tickers=TSLA&items=50&type=article&page={i}sortby=rank&days=7&token={api_key}"
    data = requests.get(api_request).json()['data']
    df = pd.DataFrame(data)
    if df.empty:
        break
    dfs.append(df)
    i += 1
tesla_df = pd.concat(dfs)

import pickle
import requests
import pandas as pd

# Get all tickers available in stocknewsapi.com
ticker_request = f"https://stocknewsapi.com/api/v1/account/tickersdbv2?token={api_key}"
json_data = requests.get(ticker_request).json()['data']
stock_news_tickers = []
for row in json_data:
    stock_news_tickers.append(row['ticker'])

# Get all Numerai Signals tickers
ticker_df = pd.read_csv("https://numerai-signals-public-data.s3-us-west-2.amazonaws.com/signals_ticker_map_w_bbg.csv")
numerai_tickers = set(ticker_df['ticker'])

# Compute intersection
relevant_tickers = list(set(stock_news_tickers).intersection(numerai_tickers))

data.loc[:, 'title'] = data['title'] + " [SEP] "

# Aggregate news headlines for each ticker
dfs = []
for ticker in relevant_tickers:
    aggregated = data[data['tickers'].apply(lambda x: ticker in x)].resample("W-fri", on='date').sum()
    aggregated = aggregated.drop("tickers", axis=1)
    aggregated['ticker'] = ticker
    aggregated = aggregated.drop_duplicates("ticker", keep='last')
    if aggregated.empty:
        continue
    dfs.append(aggregated)
new_df = pd.concat(dfs)
new_df['title'] = new_df['title'].astype(str)
merged = new_df.merge(ticker_df, on='ticker')
merged = merged.drop("yahoo", axis=1).dropna()

def _chunks(lst, n):
    """ Yield n-sized chunks from list. """
    for i in range(0, len(lst), n):
        yield lst[i:i + n]

batch_size = 8
for row in data:
    ticker_headlines = row.split(" [SEP] ")[:-1]
    .
    for batch in _chunks(ticker_headlines, batch_size):
        .

import torch
from transformers import AutoTokenizer
batch = ["The stock will go up in the upcoming week!", "Earnings are down for Q3"]
tokenizer = AutoTokenizer.from_pretrained("ipuneetrathore/bert-base-cased-finetuned-finBERT")
encoded = tokenizer(batch, add_special_tokens=True, max_length=200,
                    padding='max_length',return_attention_mask=True,
                    return_tensors='pt', truncation=True)
input_ids = torch.cat([encoded['input_ids']], dim=0).to('cuda')
attention_mask = torch.cat([encoded['attention_mask']], dim=0).to('cuda')

import torch.nn.functional as F
from transformers import AutoModelForSequenceClassification
model = AutoModelForSequenceClassification.from_pretrained("ipuneetrathore/bert-base-cased-finetuned-finBERT").eval().to('cuda')
model_output = model(input_ids, token_type_ids=None, attention_mask=attention_mask)
logits = model_output[0]
softmax_output = F.softmax(logits, dim=1).cpu().detach().numpy()

sentiment_score = softmax_output[:, 2] - softmax_output[:, 0]

mean_score = np.array(sent_scores_ticker).ravel().mean()

from sklearn.preprocessing import MinMaxScaler
def scale_sentiment(sentiments):
    """ Scale sentiment scores from [-1...1] to [0...1] """
    mm = MinMaxScaler()
    sent_proc = np.array(sentiments).reshape(-1, 1)
    return mm.fit_transform(sent_proc)

import numerapi
from datetime import datetime
from dateutil.relativedelta import relativedelta, FR

# API settings for submitting to Numerai
NMR_PUBLIC_ID = "YOUR PUBLIC KEY"
NMR_SECRET_KEY = "YOUR SECRET KEY"
MODEL_NAME = "YOUR MODEL NAME"
SUB_PATH = "finbert_submission.csv"

# Initialize API with API Keys and add data_type column
NAPI = numerapi.SignalsAPI(NMR_PUBLIC_ID, NMR_SECRET_KEY)
final_df.loc[:, "data_type"] = "live"
# Add date column denoting last Friday
friday = int(str((datetime.now() + relativedelta(weekday=FR(-1))).date()).replace("-", ""))
final_df["friday_date"] = friday

# Save final DataFrame to CSV and upload predictions
cols = ["bloomberg_ticker", "friday_date", "data_type", "signal"]
final_df[cols].reset_index(drop=True).to_csv(SUB_PATH, index=False)
model_id = NAPI.get_models()[MODEL_NAME]
NAPI.upload_predictions(SUB_PATH, model_id=model_id)