Closed livinglife2x closed 3 years ago
I checked both the zip and tar.gz source distributions. Both contain requirements.txt. I was also able to install the package using pip. I am not able to replicate your problem. Could it be that you are having a problem installing one of the dependencies? Try installing from github.
I also had this issue but was able to install per tr8dr's suggesting using:
pip install git+https://github.com/tr8dr/tseries-patterns
that said, I am having an issue trying to use data pulled from yahoo:
import yfinance as yf
from tseries_patterns import AmplitudeBasedLabeler
# download data
ticker = 'VTI'
df = yf.download(ticker, start="2000-01-01")
df = df.reset_index()
df = df.iloc[:,:-2]
df.columns = [['stamp','open','high','low','close']]
labeler = AmplitudeBasedLabeler(minamp = 10, Tinactive = 10)
labels = labeler.label(df)
labeler.plot()
this throws the error:
TypeError: only integer scalar arrays can be converted to a scalar index
any suggestions?
Hi, there is a bug in your code. When you renamed the columns you indicated a list of lists, should just be a list. Also given that this is for equities and on daily data, your minimum amplitude target should be at least 5% (or 500 bps). The inactive time should also be longer to carry through trends (say 300).
import yfinance as yf
from tseries_patterns import AmplitudeBasedLabeler
# download data
ticker = 'VTI'
df = yf.download(ticker, start="2013-01-01")
df = df.reset_index()
df = df.iloc[:,:-2]
df.columns = ['stamp','open','high','low','close']
labeler = AmplitudeBasedLabeler(minamp = 500, Tinactive = 300)
labels = labeler.label(df)
labeler.plot()
I am also adjusting the code so you do not need to rename the columns. Will check that in shortly.
thanks for the updated code - for some reason when I run it, the last line:
labeler.plot()
throws:
TypeError: groupby() got an unexpected keyword argument 'dropna'
Can you forward the error stack trace. I do not have any dropna's in the codebase. Perhaps it is in a library this depends on. Are you using a conda installation and which version? Perhaps some libraries are old (or alternatively bleeding edge).
just emailed you the error stack trace (sorry not sure what you meant specifically by forward in this case) as well as a list of all conda packages in the environment
Based on the stack trace you sent, it looks like the conda installation you are running is using an older pandas library. If you look at the documentation: [https://pandas.pydata.org/pandas-docs/stable/reference/api/pandas.DataFrame.groupby.html](pandas groupby), it supports the dropna
argument. However that support was added in pandas 1.1.0 release in July. Pandas is currently on 1.1.3.
The plotnine library (which would not have come with conda) is expecting a recent installation of pandas. Perhaps you can try updating numpy and pandas to the current version.
success!! I updated both numpy and pandas and everything is working now - thanks vm for your help in sorting this for me
I tried to install tseries_patterns with pip install tseries_patterns. Looks like requirements.txt is missing in the sdist.