llsourcell / bitcoin_prediction Goto Github PK

Do you have to make sure that tweets aren't analysed twice or more?

There should be a requirements files, lots of lib should be installed.

Hi Siraj, just wondering where the merge_data.csv comes from in the engine.py ? Its being read into a data frame but I cant see it being used anywhere else?

Should I buy more coins or not? 😄

What do you think of the sample size of 100 (and the samples twitter selects for you in your search) you get from all twitter data? Is this kind of "proof of concept" repo or did you think about that?

Great lib, great video. Thanks. Its very timely that you've realised this work.

I'm wondering, how much code did you change from Sapphirine's work? Is the goal of this lib to be a POC or are you going to maintain it and accept PRs?

Thanks

Can you see if you can validate my thoughts on being able to build a price prediction based on the number of short and long positions open on bitmex?
Check these charts out...
https://coinfarm.online/?tcs=1m

Running LSTM_V1.ipynb

data2 = pd.read_csv(filepath_or_buffer="sentiment6.txt")
data2.info()
data2.dropna(axis=1,how='all')
data2['sentiment'] = data2.sentiment.astype(float)

Results in:

AttributeError: 'DataFrame' object has no attribute 'sentiment'

sentiment6.txt starts with:
20171114,0.38021943904690936
20171113,0.38021943904690936
20171112,0.38021943904690936
20171111,0.38021943904690936
20171110,0.38021943904690936
20171109,0.38021943904690936
20171108,0.38021943904690936
...

Possible solution:
Add names to read_csv:

data1 = pd.read_csv(filepath_or_buffer="bitcoinprices.txt", names=['stamp', 'price'])
data2 = pd.read_csv(filepath_or_buffer="sentiment6.txt", names=['stamp', 'sentiment'])

Mean of all tweets? Why does the first few days have the same index?

Is it possible to add additional indicators like macd, rssi, ma (moving average) indicators to network so it could better predict movement?

Hi! Were i can find 'merged_data.csv' for ipynb usage (data = pd.read_csv("merged_data.csv"))?

I didn't find any correlation between tweeter sentiment and bitcoin prices.

Take a look at this notebook:

https://github.com/flamoedo/Bitcoin-sentiment-correlation/blob/master/Bitcoin%20daily%20returns%20correlation.ipynb

i am using python 3.6.7 , in python 3.6.7 version not able to install tensorflow, please post the solution if you have any idea.
the error meassage is shown in below,

Could not find a version that satisfies the requirement tensorflow (from versions: )
No matching distribution found for tensorflow