Multiclass Classification For Transactions
Multiclass Classification for Transactions
For this notebook we will be looking to classify a public dataset of transactions into a number of categories that we have predefined. These approaches should be replicable to any multiclass classification use case where we are trying to fit transactional data into predefined categories, and by the end of running through this you should have a few approaches for dealing with both labelled and unlabelled datasets.
The different approaches we'll be taking in this notebook are:
- Zero-shot Classification: First we'll do zero shot classification to put transactions in one of five named buckets using only a prompt for guidance
- Classification with Embeddings: Following this we'll create embeddings on a labelled dataset, and then use a traditional classification model to test their effectiveness at identifying our categories
- Fine-tuned Classification: Lastly we'll produce a fine-tuned model trained on our labelled dataset to see how this compares to the zero-shot and few-shot classification approaches
Setup
Load dataset
We're using a public transaction dataset of transactions over £25k for the Library of Scotland. The dataset has three features that we'll be using:
- Supplier: The name of the supplier
- Description: A text description of the transaction
- Value: The value of the transaction in GBP
Source:
https://data.nls.uk/data/organisational-data/transactions-over-25k/
Number of transactions: 359
Date Supplier Description \
0 21/04/2016 M & J Ballantyne Ltd George IV Bridge Work
1 26/04/2016 Private Sale Literary & Archival Items
2 30/04/2016 City Of Edinburgh Council Non Domestic Rates
3 09/05/2016 Computacenter Uk Kelvin Hall
4 09/05/2016 John Graham Construction Ltd Causewayside Refurbishment
Transaction value (£)
0 35098.0
1 30000.0
2 40800.0
3 72835.0
4 64361.0
Zero-shot Classification
We'll first assess the performance of the base models at classifying these transactions using a simple prompt. We'll provide the model with 5 categories and a catch-all of "Could not classify" for ones that it cannot place.
Transaction: M & J Ballantyne Ltd George IV Bridge Work 35098.0 Classification: Building Improvement
Our first attempt is correct, M & J Ballantyne Ltd are a house builder and the work they performed is indeed Building Improvement.
Lets expand the sample size to 25 and see how it performs, again with just a simple prompt to guide it
/var/folders/3n/79rgh27s6l7_l91b9shw0_nr0000gp/T/ipykernel_81921/2775604370.py:2: SettingWithCopyWarning: A value is trying to be set on a copy of a slice from a DataFrame. Try using .loc[row_indexer,col_indexer] = value instead See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy test_transactions['Classification'] = test_transactions.apply(lambda x: classify_transaction(x),axis=1)
Classification ,Building Improvement 17 ,Literature & Archive 3 ,Software/IT 2 ,Could not classify 2 ,Utility Bills 1 ,Name: count, dtype: int64
Initial results are pretty good even with no labelled examples! The ones that it could not classify were tougher cases with few clues as to their topic, but maybe if we clean up the labelled dataset to give more examples we can get better performance.
Classification with Embeddings
Lets create embeddings from the small set that we've classified so far - we've made a set of labelled examples by running the zero-shot classifier on 101 transactions from our dataset and manually correcting the 15 Could not classify results that we got
Create embeddings
This initial section reuses the approach from the Get_embeddings_from_dataset Notebook to create embeddings from a combined field concatenating all of our features
101
Use embeddings for classification
Now that we have our embeddings, let see if classifying these into the categories we've named gives us any more success.
For this we'll use a template from the Classification_using_embeddings notebook
precision recall f1-score support
Building Improvement 0.92 1.00 0.96 11
Literature & Archive 1.00 1.00 1.00 3
Other 0.00 0.00 0.00 1
Software/IT 1.00 1.00 1.00 1
Utility Bills 1.00 1.00 1.00 5
accuracy 0.95 21
macro avg 0.78 0.80 0.79 21
weighted avg 0.91 0.95 0.93 21
/Users/vishnu/code/openai-cookbook/cookbook_env/lib/python3.11/site-packages/sklearn/metrics/_classification.py:1565: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.
_warn_prf(average, modifier, f"{metric.capitalize()} is", len(result))
/Users/vishnu/code/openai-cookbook/cookbook_env/lib/python3.11/site-packages/sklearn/metrics/_classification.py:1565: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.
_warn_prf(average, modifier, f"{metric.capitalize()} is", len(result))
/Users/vishnu/code/openai-cookbook/cookbook_env/lib/python3.11/site-packages/sklearn/metrics/_classification.py:1565: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.
_warn_prf(average, modifier, f"{metric.capitalize()} is", len(result))
Performance for this model is pretty strong, so creating embeddings and using even a simpler classifier looks like an effective approach as well, with the zero-shot classifier helping us do the initial classification of the unlabelled dataset.
Lets take it one step further and see if a fine-tuned model trained on this same labelled datasets gives us comparable results
Fine-tuned Transaction Classification
For this use case we're going to try to improve on the few-shot classification from above by training a fine-tuned model on the same labelled set of 101 transactions and applying this fine-tuned model on group of unseen transactions
Building Fine-tuned Classifier
We'll need to do some data prep first to get our data ready. This will take the following steps:
- To prepare our training and validation sets, we'll create a set of message sequences. The first message for each will be the user prompt formatted with the details of the transaction, and the final message will be the expected classification response from the model
- Our test set will contain the initial user prompt for each transaction, along with the corresponding expected class label. We will then use the fine-tuned model to generate the actual classification for each transaction.
101
( class_id class , 0 0 Other , 1 1 Literature & Archive , 2 2 Software/IT , 3 3 Utility Bills , 4 4 Building Improvement, , 5)
Fine tuned model id: ft:gpt-4o-2024-08-06:openai::BKr3Xy8U
Applying Fine-tuned Classifier
Now we'll apply our classifier to see how it performs. We only had 31 unique observations in our training set and 8 in our validation set, so lets see how the performance is
result True 20 False 1 Name: count, dtype: int64 F1 Score: 0.9296066252587991 Raw Accuracy: 0.9523809523809523