Skip to main content

Compact Prediction Tree: A Lossless Model for Accurate Sequence Prediction

Project description


Downloads License

What is it ?

This project is a cython open-source implementation of the Compact Prediction Tree algorithm using multithreading.

CPT is a sequence prediction model. It is a highly explainable model specialized in predicting the next element of a sequence over a finite alphabet.

This implementation is based on the following research papers:


You can simply use pip install cpt.

Simple example

You can test the model with the following code:

from cpt.cpt import Cpt
model = Cpt()[['hello', 'world'],
           ['hello', 'this', 'is', 'me'],
           ['hello', 'me']

model.predict([['hello'], ['hello', 'this']])
# Output: ['me', 'is']

For an example with the compatibility with sklearn, you should check the documentation.



The model can be trained with the fit method.

If needed the model can be retrained with the same method. It adds new sequences to the model and do not remove the old ones.


The predictions are launched by default with multithreading with OpenMP.

The predictions can also be launched in a single thread with the option multithread=False in the predict method.

You can control the number of threads by setting the following environment variable OMP_NUM_THREADS.


You can pickle the model to save it, and load it later via pickle library.

from cpt.cpt import Cpt
import pickle

model = Cpt()[['hello', 'world']])

dumped = pickle.dumps(model)

unpickled_model = pickle.loads(dumped)

print(model == unpickled_model)


The CPT class has several methods to explain the predictions.

You can see which elements are considered as noise (with a low presence in sequences) with model.compute_noisy_items(noise_ratio).

You can retrieve trained sequences with model.retrieve_sequence(id).

You can find similar sequences with find_similar_sequences(sequence).

You can not yet retrieve automatically all similar sequences with the noise reduction technique.


CPT has 3 meta parameters that need to be tuned. You can check how to tune them in the documentation. To tune you can use the model_selection module from sklearn, you can find an example here on how to.


The benchmark has been made on the FIFA dataset, the data can be found on the SPMF website.

Using multithreading, CPT was able to perform around 5000 predictions per second.

Without multithreading, CPT predicted around 1650 sequences per second.

Details on the benchmark can be found here.

Further reading

A study has been made on how to reduce dataset size, and so training / testing time using PageRank on the dataset.

The study has been published in IJIKM review here. An overall performance improvement of 10-40% has been observed with this technique on the prediction time without any accuracy loss.

One of the co-author of CPT has also published an algorithm subseq for sequence prediction. An implementation can be found here

Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

cpt-1.3.0.tar.gz (113.3 kB view hashes)

Uploaded source

Built Distributions

cpt-1.3.0-cp310-cp310m-win_amd64.whl (95.0 kB view hashes)

Uploaded cp310

cpt-1.3.0-cp310-cp310m-win32.whl (81.1 kB view hashes)

Uploaded cp310

cpt-1.3.0-cp310-cp310-macosx_10_9_x86_64.whl (102.9 kB view hashes)

Uploaded cp310

cpt-1.3.0-cp39-cp39m-win_amd64.whl (95.2 kB view hashes)

Uploaded cp39

cpt-1.3.0-cp39-cp39m-win32.whl (81.1 kB view hashes)

Uploaded cp39

cpt-1.3.0-cp39-cp39-macosx_10_9_x86_64.whl (101.0 kB view hashes)

Uploaded cp39

cpt-1.3.0-cp38-cp38m-win_amd64.whl (93.2 kB view hashes)

Uploaded cp38

cpt-1.3.0-cp38-cp38m-win32.whl (76.9 kB view hashes)

Uploaded cp38

cpt-1.3.0-cp38-cp38-macosx_10_9_x86_64.whl (100.4 kB view hashes)

Uploaded cp38

cpt-1.3.0-cp37-cp37m-win_amd64.whl (91.0 kB view hashes)

Uploaded cp37

cpt-1.3.0-cp37-cp37m-win32.whl (74.9 kB view hashes)

Uploaded cp37

cpt-1.3.0-cp37-cp37m-macosx_10_9_x86_64.whl (100.9 kB view hashes)

Uploaded cp37

cpt-1.3.0-cp36-cp36m-win_amd64.whl (91.2 kB view hashes)

Uploaded cp36

cpt-1.3.0-cp36-cp36m-win32.whl (74.9 kB view hashes)

Uploaded cp36

cpt-1.3.0-cp36-cp36m-macosx_10_9_x86_64.whl (103.6 kB view hashes)

Uploaded cp36

Supported by

AWS AWS Cloud computing Datadog Datadog Monitoring Facebook / Instagram Facebook / Instagram PSF Sponsor Fastly Fastly CDN Google Google Object Storage and Download Analytics Huawei Huawei PSF Sponsor Microsoft Microsoft PSF Sponsor NVIDIA NVIDIA PSF Sponsor Pingdom Pingdom Monitoring Salesforce Salesforce PSF Sponsor Sentry Sentry Error logging StatusPage StatusPage Status page