File size: 5,239 Bytes
d24798b
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
0b929da
d24798b
 
 
 
0b929da
d24798b
0b929da
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
d24798b
0b929da
 
d24798b
 
 
 
 
0b929da
d24798b
0b929da
 
d24798b
 
 
 
 
 
 
0b929da
d24798b
 
 
 
 
 
 
 
 
0b929da
d24798b
0b929da
 
 
 
 
d24798b
 
 
 
0b929da
d24798b
 
 
 
 
 
 
0b929da
 
d24798b
0b929da
 
d24798b
0b929da
 
d24798b
 
 
 
 
 
 
 
 
 
 
 
0b929da
 
 
 
 
d24798b
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
import os
import datetime as dt
import pandas as pd
import torch
import gradio as gr
import yfinance as yf

from chronos import BaseChronosPipeline  # from 'chronos-forecasting'

# ---- ์ „์—ญ ์บ์‹œ: ๋ชจ๋ธ์„ ํ•œ ๋ฒˆ๋งŒ ๋กœ๋“œํ•ด ์žฌ์‚ฌ์šฉ ----
_PIPELINE_CACHE = {}

def get_pipeline(model_id: str, device: str = "cpu"):
    key = (model_id, device)
    if key not in _PIPELINE_CACHE:
        _PIPELINE_CACHE[key] = BaseChronosPipeline.from_pretrained(
            model_id,
            device_map=device,  # "cpu" / "cuda"
            torch_dtype=torch.float32 if device == "cpu" else torch.bfloat16,
        )
    return _PIPELINE_CACHE[key]

# ---- ์ฃผ๊ฐ€/ํฌ๋ฆฝํ†  ๋ฐ์ดํ„ฐ ๋กœ๋”ฉ (yfinance, ๊ฒฌ๊ณ ํ™”) ----
def load_close_series(ticker: str, start: str, end: str, interval: str = "1d"):
    """
    BTC-USD ๋“ฑ ํฌ๋ฆฝํ†  ์‹ฌ๋ณผ์—์„œ ๊ฐ„ํ—์ ์œผ๋กœ timezone/ํŒŒ์‹ฑ ์˜ค๋ฅ˜๊ฐ€ ๋‚˜๋ฏ€๋กœ
    history() ๊ฒฝ๋กœ๋ฅผ ์šฐ์„  ์‚ฌ์šฉํ•˜๊ณ , ์‹คํŒจ ์‹œ ํ•œ ๋ฒˆ ์žฌ์‹œ๋„.
    """
    # ๊ธฐ๋ณธ๊ฐ’ ๋ณด์ •: ๋„ˆ๋ฌด ์ตœ๊ทผ๋งŒ ๊ณ ๋ฅด๋ฉด ๋นˆ ๋ฐ์ดํ„ฐ๊ฐ€ ๋‚˜์˜ฌ ์ˆ˜ ์žˆ์–ด ์ผ๋ด‰์€ ๊ณผ๊ฑฐ๋ถ€ํ„ฐ ๊ถŒ์žฅ
    _start = start or "2014-09-17"  # BTC-USD ์ตœ์ดˆ ์ƒ์žฅ์ผ ๊ทผ์ฒ˜
    _end = end or dt.date.today().isoformat()

    tk = yf.Ticker(ticker)
    try:
        df = tk.history(start=_start, end=_end, interval=interval, auto_adjust=True, actions=False)
        if df.empty or "Close" not in df:
            raise ValueError("empty")
    except Exception:
        # fallback: download() ๊ฒฝ๋กœ ์‹œ๋„
        df = yf.download(ticker, start=_start, end=_end, interval=interval, progress=False, threads=False)
        if df.empty or "Close" not in df:
            raise ValueError("๋ฐ์ดํ„ฐ๊ฐ€ ์—†๊ฑฐ๋‚˜ 'Close' ์—ด์ด ์—†์Šต๋‹ˆ๋‹ค. ํ‹ฐ์ปค/๋‚ ์งœ/๊ฐ„๊ฒฉ์„ ํ™•์ธํ•˜์„ธ์š”.")

    s = df["Close"].dropna().astype(float)
    if s.empty:
        raise ValueError("๋‹ค์šด๋กœ๋“œ ๊ฒฐ๊ณผ๊ฐ€ ๋น„์–ด ์žˆ์Šต๋‹ˆ๋‹ค. ๊ธฐ๊ฐ„/๊ฐ„๊ฒฉ์„ ์ค„์ด๊ฑฐ๋‚˜ ๋‹ค์‹œ ์‹œ๋„ํ•˜์„ธ์š”.")
    return s

# ---- ์˜ˆ์ธก ํ•จ์ˆ˜ (Gradio๊ฐ€ ํ˜ธ์ถœ) ----
def run_forecast(ticker, start_date, end_date, horizon, model_id, device, interval):
    try:
        series = load_close_series(ticker.strip(), start_date, end_date, interval)
    except Exception as e:
        # Gradio v4์—์„œ๋Š” Plot.update๊ฐ€ ์—†์Œ โ†’ None ๋ฐ˜ํ™˜์œผ๋กœ ์ •๋ฆฌ
        return None, pd.DataFrame(), f"๋ฐ์ดํ„ฐ ๋กœ๋”ฉ ์˜ค๋ฅ˜: {e}"

    pipe = get_pipeline(model_id, device)
    H = int(horizon)

    # Chronos ์ž…๋ ฅ: 1D ํ…์„œ (float)
    context = torch.tensor(series.values, dtype=torch.float32)

    # ์˜ˆ์ธก: (num_series=1, num_quantiles=3, H) with q=[0.1, 0.5, 0.9]
    preds = pipe.predict(context=context, prediction_length=H)[0]
    q10, q50, q90 = preds[0], preds[1], preds[2]

    # ํ‘œ ๋ฐ์ดํ„ฐ
    df_fcst = pd.DataFrame(
        {"q10": q10.numpy(), "q50": q50.numpy(), "q90": q90.numpy()},
        index=pd.RangeIndex(1, H + 1, name="step"),
    )

    # ๋ฏธ๋ž˜ x์ถ•: interval์— ๋งž๋Š” pandas ์ฃผ๊ธฐ
    import matplotlib.pyplot as plt
    freq_map = {"1d": "D", "1h": "H", "30m": "30T", "15m": "15T", "5m": "5T"}
    freq = freq_map.get(interval, "D")
    future_index = pd.date_range(series.index[-1], periods=H + 1, freq=freq)[1:]

    # ๊ทธ๋ž˜ํ”„
    fig = plt.figure(figsize=(10, 4))
    plt.plot(series.index, series.values, label="history")
    plt.plot(future_index, q50.numpy(), label="forecast(q50)")
    plt.fill_between(future_index, q10.numpy(), q90.numpy(), alpha=0.2, label="q10โ€“q90")
    plt.title(f"{ticker} forecast by Chronos-Bolt ({interval}, H={H})")
    plt.legend()
    plt.tight_layout()

    note = "โ€ป ๋ฐ๋ชจ ๋ชฉ์ ์ž…๋‹ˆ๋‹ค. ํˆฌ์ž ํŒ๋‹จ์˜ ์ฑ…์ž„์€ ๋ณธ์ธ์—๊ฒŒ ์žˆ์Šต๋‹ˆ๋‹ค."
    return fig, df_fcst, note

# ---- Gradio UI ----
with gr.Blocks(title="Chronos Stock/Crypto Forecast") as demo:
    gr.Markdown("# Chronos ์ฃผ๊ฐ€ยทํฌ๋ฆฝํ†  ์˜ˆ์ธก ๋ฐ๋ชจ")
    with gr.Row():
        ticker = gr.Textbox(value="BTC-USD", label="ํ‹ฐ์ปค (์˜ˆ: AAPL, MSFT, 005930.KS, BTC-USD)")
        horizon = gr.Slider(5, 365, value=90, step=1, label="์˜ˆ์ธก ์Šคํ… H (๊ฐ„๊ฒฉ ๋‹จ์œ„์™€ ๋™์ผ)")
    with gr.Row():
        start = gr.Textbox(value="2014-09-17", label="์‹œ์ž‘์ผ (YYYY-MM-DD, ์˜ˆ: 2014-09-17)")
        end = gr.Textbox(value=dt.date.today().isoformat(), label="์ข…๋ฃŒ์ผ (YYYY-MM-DD, ๋น„์›Œ๋‘๋ฉด ์˜ค๋Š˜)")
    with gr.Row():
        model_id = gr.Dropdown(
            choices=[
                "amazon/chronos-bolt-tiny",
                "amazon/chronos-bolt-mini",
                "amazon/chronos-bolt-small",
                "amazon/chronos-bolt-base",
            ],
            value="amazon/chronos-bolt-small",
            label="๋ชจ๋ธ"
        )
        device = gr.Dropdown(choices=["cpu"], value="cpu", label="๋””๋ฐ”์ด์Šค")
        interval = gr.Dropdown(
            choices=["1d", "1h", "30m", "15m", "5m"],
            value="1d",
            label="๊ฐ„๊ฒฉ"
        )
    btn = gr.Button("์˜ˆ์ธก ์‹คํ–‰")

    plot = gr.Plot(label="History + Forecast")
    table = gr.Dataframe(label="์˜ˆ์ธก ๊ฒฐ๊ณผ (๋ถ„์œ„์ˆ˜)")
    note = gr.Markdown()

    btn.click(
        fn=run_forecast,
        inputs=[ticker, start, end, horizon, model_id, device, interval],
        outputs=[plot, table, note]
    )

if __name__ == "__main__":
    demo.launch()