Machine learning for your business

Ämne: Avancerad Analys

With all the hype around big data, machine learning, and data science, it’s difficult to see how and where these concepts give your business the advantage. This blog post is to help you discover and imagine ways to take advantage of machine learning, and data mining in general. It isn’t about how to do machine learning.

Please read on if you want to discover how to leverage machine learning in your business.

The 2 ways of Machine learning

Data scientists help a business leverage data to make decisions. One tool we use to do this is machine learning. At BizOne, we categorize machine learning tasks for business decisions into one of two working pipelines. We call the two pipelines ’lab work’ and ’operations work’. For ’lab work’ we apply machine learning to provide decision support to the business, and for ’operations work’ we apply machine learning to deliver decisions within operational systems. In both cases the end results are to make decisions that improve metrics.

Machine learning in the lab:

  1. Starts from a question
  2. Data scientist applies machine learning to answer or help answer the question
  3. Output is in the form of a report or scoring in a data base
  4. A business decision is made using the output
  5. Action is taken based on the decision

Machine learning in an operational system:

  1. Starts with new incoming information
  2. Operational system applies machine learning
  3. A customer facing decision is made

Applying machine learning in an operational system takes some development upfront too. It’s typically the case that we do a some lab work first that leads to the models used in the operational system.

Machine Learning Pipelines

Lab Work

  • Driven by Questions
  • Fixed Data
  • Ad-hoc, post-hoc
  • Output report or database scoring

Operations Work

  • Metric-driven
  • Automated
  • Systematic
  • Output customer-facing decisions

Example

We can explore these two pipelines with a toy webpage example. Assume the wireframe below is part of a signup flow for a web app we are running. There are many decisions left to be made for this page. Here are three decisions:

  1. Image to use, currently grey scale gift box
  2. Upgrade cost, currently XX.xx
  3. Offer valid for time, currently hh:mm:ss

One last thing, we need to make the assumption that we already have some data to keep this post short.

Wireframe(1)

Lets focus in on the decision number 1 from the list above, about the image of a gift box.

Lab Work Steps

  1. The Question is which color should the gift box be?
  2. We apply some machine learning or statistics to compare the performance of the different colors in the past in different locations and times of the day. Or, we can design an experiment to try multiple colors.
  3. Outcomes are recorded in a report.
  4. The product owner makes a color choice based on the report.
  5. Web developers change the color to the one selected by the product owner to display for the right location and time of day.

Operational system Steps

  1. A person from Stockholm is signing up for the web app at 17:50 on a early December evening.
  2. Machine learning model with in the product makes a decision, and thus displays the most likely color for the gift box to increase the chances of this visitor selecting the upgrade offer.
  3. Number of upgrades during signup increases.

Now, how would you apply this methodology to the other two decisions in the list above?

In this post we haven’t explained what machine learning is, but hopefully you’ve been exposed to how it can fit inside the decision process for your business.