References you could read before coming :

NA

Material required :

Computer
Python 3.7
Spyder or PyCharm IDE
Libraries will be installed during the training. Just in case:
· "PyPDF2"
· "nltk"
· "pandas"
· "gensim"
· "sklearn"
· "numpy"

Abstract:

Gathering unstructured data is becoming a hot topics for insurance and banking firms. Such data help companies to improve their risk assessment, but also to improve their understanding of market needs, of the competition and to identify new claims. For example in the insurance industry, collecting information related to insurance products competitors is a strategic element. But how to collect these information from the web ? How to handle both quantitative and qualitative information in documents to get structured data ?

Goal:

In this business case, we will present how to collect Belgium insurance general terms and conditions from the web and how to make the best of these information using Python. We will develop text mining and machine learning models to identify paragraphs topics and patterns within documents. We will also study regular expressions to extract key criteria and amounts in order to create benchmarks.

Agenda :

1. Introduction and context
2. File collection: Import of general terms and conditions in pdf file
3. Data management : data cleaning (stop words, symbols, lemmatization, POS), pre-processing, etc.
4. Introductive analysis: N-grams analysis, barplots frequency, wordclouds, multivariate analysis
5. Regex: Regex functions, Python syntax, examples with email address and URL
6. Occurrences based approach: Term Document Matrix and distance measure, heatmap data visualization, unsupervised machine learning example with HAC, dendrograms, PCA and main axis of study, k-means comparison.
7. Word2Vec based approach : Word embedding, impact on HAC and K-mean techniques.

Practicalities

• Lunch is included in your ticket.

• Doors are open at 8.30 am.

• Training starts at 9 am and finishes at 5 pm
  (be on time or inform us if delay, respect for the audience/trainer).

• Accessibility: public transport (stop Yser / Ijzer - subway line 2-6 ).
  We do not have parking but normally there are some parking slots at Rue des Commerçants.

• Training partners and community can contact us via e-mail to request their discount code (training@di-academy.com).

// If you want to avoid the EventBrite fee and pay directly to us, drop a message confirming your attendance and providing your full details (company name, address, TVA # and # of seats) to training@di-academy.com asap //