Implement model

.env

@@ -0,0 +1,2 @@
+SCALER_MAX_VALUE=10.589996337890625
+SCALER_MAX_VOLUME=1460852400.0

.gitattributes

@@ -32,3 +32,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+*.data* filter=lfs diff=lfs merge=lfs -text

.gitignore

@@ -1 +1,3 @@
-env
+env
+.DS_Store
+test.py

app.py

@@ -1,9 +1,50 @@
+from dotenv import load_dotenv
+load_dotenv('.env')
+
 import gradio as gr
+from utils import *
+import os
+
+
+days_to_plot = 50
+
+data = get_data().iloc[-500:]
+
+
+data_to_plot = data.iloc[-days_to_plot:][["Close"]]
+data_to_plot['date'] = data_to_plot.index
 
+with gr.Blocks() as demo:
+    gr.Markdown("# Apple Predictor")
+    predict_button = gr.Button("Predict")
+    with gr.Row() as row0:
+        with gr.Column() as col0:
+            gr.Markdown("## Last candle info")
+            last_open = gr.Textbox(get_last_candle_value(data, 'Open') ,label="Last Open")
+            last_max = gr.Textbox( get_last_candle_value(data, 'High') ,label="Last Max")
+            last_min = gr.Textbox( get_last_candle_value(data, 'Low') ,label="Last Min")
+            last_close = gr.Textbox( get_last_candle_value(data, 'Close') ,label="Last Close")
 
-def greet(name):
-    return "Hello negrito " + name + "!!"
+        with gr.Column() as col1:
+            gr.Markdown("## Next Candle Prediction")
+            jump_text = gr.Textbox(label="Jump")
+            open_text = gr.Textbox(label="Open")
+            max_text = gr.Textbox(label="Max")
+            min_text = gr.Textbox(label="Min")
+            next_close_text = gr.Textbox(label="Close")
+    with gr.Row() as row1:
+        value_plot = gr.LinePlot(data_to_plot,
+                                 x="date",
+                                 y="Close",
+                                 label=f'Last {days_to_plot} days',
+                                 y_lim=[float(data_to_plot['Close'].min())-5, float(data_to_plot['Close'].max())+5])
 
+    outputs = [jump_text,
+               open_text,
+               max_text,
+               min_text,
+               next_close_text
+               ]
+    predict_button.click(lambda: predict(data), outputs=outputs)
 
-iface = gr.Interface(fn=greet, inputs="text", outputs="text")
-iface.launch()
+demo.launch(debug=True)

requirements.txt

@@ -0,0 +1,79 @@
+aiofiles==23.1.0
+aiohttp==3.8.4
+aiosignal==1.3.1
+altair==4.2.2
+anyio==3.6.2
+appdirs==1.4.4
+async-timeout==4.0.2
+attrs==22.2.0
+beautifulsoup4==4.12.2
+certifi==2022.12.7
+cffi==1.15.1
+charset-normalizer==3.1.0
+click==8.1.3
+contourpy==1.0.7
+cryptography==40.0.1
+cycler==0.11.0
+entrypoints==0.4
+fastapi==0.95.0
+ffmpy==0.3.0
+filelock==3.11.0
+fonttools==4.39.3
+frozendict==2.3.7
+frozenlist==1.3.3
+fsspec==2023.4.0
+gradio==3.24.1
+gradio_client==0.0.8
+h11==0.14.0
+html5lib==1.1
+httpcore==0.16.3
+httpx==0.23.3
+huggingface-hub==0.13.4
+idna==3.4
+importlib-resources==5.12.0
+Jinja2==3.1.2
+jsonschema==4.17.3
+kiwisolver==1.4.4
+linkify-it-py==2.0.0
+lxml==4.9.2
+markdown-it-py==2.2.0
+MarkupSafe==2.1.2
+matplotlib==3.7.1
+mdit-py-plugins==0.3.3
+mdurl==0.1.2
+multidict==6.0.4
+multitasking==0.0.11
+numpy==1.24.2
+orjson==3.8.10
+packaging==23.0
+pandas==2.0.0
+pandas-datareader==0.10.0
+Pillow==9.5.0
+pycparser==2.21
+pydantic==1.10.7
+pydub==0.25.1
+pyparsing==3.0.9
+pyrsistent==0.19.3
+python-dateutil==2.8.2
+python-multipart==0.0.6
+pytz==2023.3
+PyYAML==6.0
+requests==2.28.2
+rfc3986==1.5.0
+semantic-version==2.10.0
+six==1.16.0
+sniffio==1.3.0
+soupsieve==2.4
+starlette==0.26.1
+toolz==0.12.0
+tqdm==4.65.0
+typing_extensions==4.5.0
+tzdata==2023.3
+uc-micro-py==1.0.1
+urllib3==1.26.15
+uvicorn==0.21.1
+webencodings==0.5.1
+websockets==11.0.1
+yarl==1.8.2
+yfinance==0.2.16
+zipp==3.15.0

utils.py

@@ -0,0 +1,97 @@
+import numpy as np
+# import matplotlib.pyplot as plt
+import pandas as pd
+import pandas_datareader as web
+import datetime as dt
+import yfinance as yfin
+# import tensorflow as tf
+import os
+# import re
+
+from huggingface_hub import from_pretrained_keras
+# from sklearn.preprocessing import MinMaxScaler
+# from tensorflow.keras.models import Sequential
+# from tensorflow.keras.layers import Dense, Dropout, LSTM
+
+
+def get_data(ticker='AAPL', start=None, end=None):
+    if end is None:
+        end = dt.date.today()
+    if start is None:
+        start = end - dt.timedelta(days=800)
+
+    yfin.pdr_override()
+    data = web.data.get_data_yahoo(ticker, start, end)
+    # data = pd.read_csv('train_data.csv', index_col='Date')
+    return data
+
+
+def get_last_candle_value(data, column):
+    val = data.iloc[-1][column]
+    return "{:.2f}".format(val)
+
+
+# Preprocessing functions copied from notebook where model was trained
+def create_remove_columns(data):
+    # create jump column
+    data = pd.DataFrame.copy(data)
+    data['Jump'] = data['Open'] - data['Close'].shift(1)
+    data['Jump'].fillna(0, inplace=True)
+    data.insert(0,'Jump', data.pop('Jump'))
+    return data
+
+def normalize_data(data):
+    # Returns a tuple with the normalized data, the scaler and the decoder
+    # The normalized data is a dataframe with the following columns:
+    # ['Jump', 'High', 'Low', 'Close', 'Adj Close', 'Volume']
+    the_data = pd.DataFrame.copy(data)
+    # substract the open value to all columns but the first one and the last one which are "Jump" and "Volume"
+    the_data.iloc[:, 1:-1] = the_data.iloc[:,1:-1] - the_data['Open'].values[:, np.newaxis]
+    # print('the_data')
+    # print(the_data)
+
+    the_data.pop('Open')
+    # Create the scaler
+    max_value = float(os.getenv('SCALER_MAX_VALUE'))
+    max_volume = float(os.getenv('SCALER_MAX_VOLUME'))
+    def scaler(d):
+        data = pd.DataFrame.copy(d)
+        print('max_value: ', max_value)
+        print('max_volume: ', max_volume)
+        data.iloc[:, :-1] = data.iloc[:,:-1].apply(lambda x: x/max_value)
+        data.iloc[:, -1] = data.iloc[:,-1].apply(lambda x: x/max_volume)
+        return data
+    def decoder(values):
+        decoded_values = values * max_value
+        return decoded_values
+    
+    normalized_data = scaler(the_data)
+
+    return normalized_data, scaler, decoder
+
+def preprocessing(data):
+    # print(data.head(3))
+    data_0 = create_remove_columns(data)
+    # print(data_0.head(3))
+    #todo: save the_scaler somehow to use in new runtimes
+    norm_data, scaler, decoder = normalize_data(data_0)
+    # print(norm_data.head(3))
+    # print(x_train.shape, y_train.shape)
+    norm_data_array = np.array(norm_data)
+    return np.expand_dims(norm_data_array, axis=0), decoder
+
+
+# Model prediction
+# model = from_pretrained_keras("jsebdev/apple_stock_predictor")
+def predict(data):
+    input, decoder = preprocessing(data)
+    print("input")
+    print(input.shape)
+    result = decoder(model.predict(input))
+    last_close = data.iloc[-1]['Close']
+    next_candle = result[0, -1]
+    print('next_candle')
+    print(next_candle)
+    jump = next_candle[0]
+    next_candle = next_candle + last_close
+    return (jump, next_candle[0], next_candle[1], next_candle[2], next_candle[3])