aj-cloete / pssa

Singular Spectrum Analysis for time series forecasting in Python

pySSA

Singular Spectrum Analysis for time series forecasting in Python

An example of the implementation of this code can be found in Singular Spectrum Analysis Example.ipynb.

I will update the list of references/credits at another time. The research of Nina Golyandina from Russia was invaluable in aiding my understanding of this method.

[update 4 April 2017] - Please note that the current version requires a pandas dataframe with a datetime formatted index for it to work. I will push an update soon to allow numpy array inputs.

Please feel free to fork the project and contribute!

pssa's Issues

What is the best way to install this package?

Since there is no setup.py, there is no way to pip install this package even using git+https://github.com/aj-cloete/pySSA.git. Do I just clone and copy the files, or is there a better way?

forecast_recurrent error

Hello!
I like your lib! But I've got some errors:

AttributeError Traceback (most recent call last)
~/Documents/notebooks/vk/SSA/mySSA.py in forecast_recurrent(self, steps_ahead, singular_values, plot, return_df, **plotargs)
206 try:
--> 207 self.X_com_hat
208 except(AttributeError):

AttributeError: 'mySSA' object has no attribute 'X_com_hat'

During handling of the above exception, another exception occurred:

KeyError Traceback (most recent call last)
~/Documents/notebooks/vk/SSA/mySSA.py in _forecast_prep(self, singular_values)
185 for i in singular_values:
--> 186 self.forecast_orthonormal_base[i] = self.orthonormal_base[i]
187 except:

KeyError: 3

During handling of the above exception, another exception occurred:

TypeError Traceback (most recent call last)
in ()
----> 1 ssa.forecast_recurrent(steps_ahead=ssa.ts.shape[0], singular_values=streams10, plot=True)

~/Documents/notebooks/vk/SSA/mySSA.py in forecast_recurrent(self, steps_ahead, singular_values, plot, return_df, **plotargs)
207 self.X_com_hat
208 except(AttributeError):
--> 209 self._forecast_prep(singular_values)
210 self.ts_forecast = np.array(self.ts_v[0])
211 for i in range(1, self.ts_N+steps_ahead):

~/Documents/notebooks/vk/SSA/mySSA.py in _forecast_prep(self, singular_values)
189 self.forecast_orthonormal_base[0] = self.orthonormal_base[0]
190 else:
--> 191 raise('Please pass in a list/array of singular value indices to use for forecast')
192 else:
193 self.forecast_orthonormal_base = self.orthonormal_base

TypeError: exceptions must derive from BaseException

Filling gaps using SSA

Hey aj, how are you?

I'm going to use your code to fill gaps in sea level data. How should I proceed? Use this method to my entire series and then extract correspondent values for each gap or there is another way to do this?

'm' is not defined

Hello,

I am trying to learn to use SSA and I wanted to use your tutorial to see how to use it and how ti works. But when I tried to use your code I get an error message :

" File "...\pssa-master\src\mySSA.py", line 91, in embed
self.X = m(linalg.hankel(self.ts, np.zeros(self.embedding_dimension))).T[:,:self.K]
NameError: name 'm' is not defined"

In mySSA, in embed, in several lines, there is a 'm', that is not defined. Could you tell me how to defined it ? Or at least explain what it does please ?

Thanks in advance

forecast_recurrent Error

Hey aj, how are you?

Trying to use ssa.forecast_recurrent() with my data I'm getting the message: "TypeError: exceptions must derive from BaseException." and if I run again my script, I'm getting: "AttributeError: 'mySSA' object has no attribute 'R'".

Do you have any idea about what are they?

Thanks!