Time series excercises for the ACM data science summer school - Athens 2018
These exercises are designed to provide a review of some common AR modelling commands in python and a brief look at time series classification. All exercises are contained in the notebook (time-series.ipynb
) but note that submissions are required in a separate .txt or .pdf file (not .doc[x]?).
We assume that Python 3 + Anaconda are installed; furthermore the following packages are required:
- numpy
- pandas
- statsmodels
- matplotlib
- scikit-learn
- jupyter
Have fun!
-
I don't have Anaconda/Python3. We highly recommend installing Anaconda. See installation instructions here. Miniconda is also a good option if you're comfortable using the command line. You're welcome to use Python without Anaconda, but you're on your own if you do!
-
I don't have the above packages / I'm not sure I have the above packages. It's often sensible to set up a virtual environment for new package requirements.
- I know what a virtual environment is, and no, I don't want a new one. Sure, activate an existing virtual environment and install the above packages using
conda install <package name(s)>
. - I don't know what a virtual environment is, or I want to create a new one. The following command should generate an environment called
timeseries
with the required packages:
- I know what a virtual environment is, and no, I don't want a new one. Sure, activate an existing virtual environment and install the above packages using
conda create -n timeseries python=3 matplotlib numpy pandas statsmodels scikit-learn jupyter
- How do I run the notebook?
Let's call the virtual environment in step 2
<env name>
. Jupyter can then be launched from either:- Anaconda Navigator: please select
<env name>
from the dropdown menu at the top ("applications on..." or similar) first, and then launch jupyter notebook; or - Command line: by activating the environment and then launching jupyter. This can be achieved with the following two lines at the command line:
- Anaconda Navigator: please select
conda activate <env name>
jupyter notebook
The notebook server should appear in your browser. Simply navigate to the notebook file and open it.