• theano
• pandas
• numpy
• matplotlib
• sklearn
• scikit-learn
• tensorflow
• keras

Notes:

• Primary installs listed at the end of this file
• All historical data columns need to be flipped vertically to start with earliest date, down to latest date (how to flip at end of doc)
• virtual_env: workon crypto

Selected Tickers:

Quandl Key: EfVSTGzAz3sxDyyG2Tqm Aplhavantage Key: PBRGXKAUD9LKYI3Z

Note: - Needed to sym-link alpha-vantage to virtualenv crypto to be able to import package:

cd ~/.virtualenvs/crypto/lib/python3.6/site-packages/
ln -s /usr/local/lib/python3.6/dist-packages/alpha_vantage alpha_vantage

• rnnQCOM.py - Same basic functionality as the above Crypto version, with adaptations for Stock/Exchange currencies...
• rnnTSLA.py - same as above but for Tesla

####API Links:

• Quandl
• AlphVantage

####Historical Ticker Data:

• TESLA:

  • YAHOO
  • Nasdaq

• QUALCOM:

  • YAHOO
  • Nasdaq

Using Bitstamp data Missing Data Historical Data Quandl BTC API Codes Get Bitcoin CSV Data

• rnn.py
  • Gets trainging data 'csv' file and checks the size of column 2 (Opening Price)
  • Using the size of the column it trains the LSTM up to the last value stored
  • Make URL request for yesterdays Bitcoin data and extracts the latest 'Open' value
  • Uses this value as the "real_stock_price" for the prediction method
  • Appends last 'Open' price and new Predicted value to text file
  • Appends last 'Open' value to csv column 2 for tomorrow's prediction/evaluation
• rnn[2018] BACKUP1.py - BTCUSD crypto predictor. Graphs test set over predicted values for 30 days of BTC opening price
• rnn[2018] BACKUP2.py - ? (possibly same as above)
• rnn[2018] BACKUP3.py - Uses full dataset (no test set removed) to predict one days value (opening price)

Various API's to retrieve crypto data

• api1.py - Downloads Yahoo Financial data and graphs it
• api2.py - Using Quandl to retrieve bitcoin (BITSTAMP) data and then graph the 'Low' values from that data (Can't get Open price?)
• api3.py - URL request data retrieval, returns byte type then converts to dictionary and extracts opening bitcoin value
• api4.py
  • URL request data retrieval returns byte type then converts to dictionary and extracts opening bitcoin value.
  • Checks for 'open.txt' file. If doesn't exist, creates it and writes to it.
  • If it exists, appends new Opening value with date.
• api5.py - All of the above and also appends entire dictionary content to csv file under correct columns

Various API's to retrieve stock data

• api1_yahoo.py - Using Yahoo API to request for Qualcomm/Tesla stock data (Discontinued)

• api2_qcom.py

  • Using Quandl & Stock's "Ticker Symbol" to retrieve the stock data and then plot the 'High' values
  • Also prints first & last 3 lines of the data

• api3_qcom.py

  • URL request data retrieval from Yahoo Finance
  • Returns and prints date / Close price since beginning of the year

• api4_quandl.py - Using Quandl API to request for Qualcomm/Tesla stock data

• api6_quandl.py

  • Using yahoo_fin package to fetch financial data, extract opening price
  • NOT FETCHING TODAYS DATA, ONLY 2 DAYS AGO???

Various CSV file control testing

• csv_test1.py - Reads values from column 2, print number of values, append date and 'string' to last row of first 2 columns
• csv_test2.py
  • Gets number of rows in column 2 = size
  • Prints specified cell according to [col, row] (Note: Cells start at [0, 0])
  • Gets yesterdays Opening & Direction
  • Indicates Y/N for direction correct
  • Shows Error Rate from yesterday to todays Opening price
• csv_test3.py - not sure?

Github prediction project (untested)

Flip_Data_Columns:

• Using LibreOffice Writer
• Hightlight all data (excluding the headings)
• Data -> Descending

Useful Stock Tickers:

• MSFT - Microsoft
• QCOM - Qualcomm
• TSLA - Tesla
• AAPL - Apple

Tested on Google stock price and bitcoin opening price (edited from original - I think)

Edited and tested - rnn.py example from udemy using BTCUSD Test and Training set + post processing graphic visualisation
3 differnt versions of the actual RNN (best version: rnn.py?)

###Dependenies to pip3 install:

 | six  			| 1.9.0		| 1.9.0
 | scipy  			| 0.11		| 0.14
 |  numpy  		| 1.7.1 		| 1.9.1

tensorflow | | 1.0.0 | 1.2.1 | six | 1.10.0 | 1.10.0 | wheel | 0.26 | 0.26 | setuptools | setuptools | xx keras | | 2.0.1 | 2.0.6 | theano | xx | | pyyaml | xx | | six | xx | | numpy | xx | | scipy | xx |