F5 Networks, Inc. is a transnational company that specializes in application services and application delivery networking (ADN). F5 technologies focus on the delivery, security, performance, and availability of web applications, including the availability of computing, storage, and network resources.
The F5 Customer Engagement Center (CEC) is responsible for hosting briefings and presentations to potential customers to highlight what F5 can do to help their business.
The CEC team is looking to quantify their business impact which is defensible for budget justification, expansion etc. and use the data to better plan future briefings.
Akvelon provided both data science and automation support for an interactive CEC business impact dashboard.
The analysis consisted of connecting CEC briefings with product and service sales by means of customer matching and consideration of the content/type of the briefing and product sold.
Akvelon worked closely with the F5 team to establish a consistent common set of bridging categories for CEC briefings and products. Natural Language Processing (Latent Semantic Analysis) was then used to parse/vectorize raw text of briefing content before classified into several product categories with a Random Forest model.
Python, Data Science, Machine Learning, Natural Language Processing, Pandas, NumPy, Scikit-Learn, Latent Semantic Analysis, Random Forest Classification, gensim
Akvelon delivered on the attribution analysis by means of automating the machine learning data science pipeline to create the interactive dashboard. The benefits of the project include:
- Automatic generation of dashboard means a daily up-to-date look at the CEC business impact
- The advanced natural language processing and machine learning leveraged the maximum amount of signal present in the clients data
- Generated defensible analysis for CEC team to use for budget justification, expansion etc.
- Identified ways for possible CEC proactive marketing efforts to maximize impact
- Provided several filters and parameters which can be tweaked at the user’s discretion for a concentrated customized local analysis