Lessons Learned After 16 Weeks of Data Analytics

The film Minority Report starring Tom Cruise as John Anderton, was released in theaters way back in 2002 but was loosely based on a 1956 short story “The Minority Report” by Philip K. Dick (Hollywood Reporter).

It’s set in the future or the year 2054 and there is literally no crime in Washington D.C. because crimes are predicted and enforced beforehand. John Anderton heads the bureau, and he religiously believes in the system.

Like any good storyline, things take a turn for the worse when there is speculation over whether John has committed a future crime, as he is predicted to have done so.

If you’d seen the movie, you’d recognize and appreciate that there is a focus on predictive and integrated models.

Now that the Finance Transformation and Data Analytics Certificate Program has come to an end, I’ve learned so much more about predictive analytics and so the potential for finance to add value is fascinating.

According to Jim DeLoach, CFO Network, Forbes Magazine 2021, “CFOs must recognize that continual advancements in their analytics capabilities are now table stakes for the board of directors, CEOs, their executive management peers, and the many internal customers within their organizations…. It’s time for CFOs to step up and get comfortable with a new era in finance for advanced data analytics.”

It’s a bit much to summarize everything that I learned from the past 16 weeks of classes and projects, so I’ll mention just a few questions and answers which I hope are of value to you.

Question 1: What are the different stages involved in a typical journey for the Analytics Business Partner (Finance Business Partner)?

Answer: The stages involved are:

Stage 1 - Providing timely and accurate information (Hindsight)

Stage 2 – Eliminating the ineffective which frees up time, providing better information using centers of excellence, automation, and data visualization (Efficient)

Stage 3 – Providing deeper information (Insight)

Stage 4 - Providing predictive analytics to drive impact (Foresight).

Stage 4 requires a different way of thinking and so the analytics mindset can answer all of the questions below:

What happened? E.g., Sales fell by 15% between last year and this year (YOY) (Tools - Excel, dashboards, BI, EPM, etc.)

Where it happened? In the Asian Export Market. (Tools - Excel, dashboards, BI, EPM, etc.)

Why it happened? The market is price sensitive and very attractive and so increased competition impacted sales.?(Tools - Analytics, Stats, AI, ML, etc.)

What will happen? Sales will trend upwards but by 15% against budget for the rest of the year or the next 6 months (tools - Analytics, Stats, AI, ML, etc.)

How to make it happen? To be achieved by promotions to promote our brand based on quality and not on price to avoid price wars. Our seasonal nature will also improve sales within this period when compared to the first half of the year based on trends and a new product, P will also hit the Asian Export Market next month and is expected to bring in X amount of sales (Tools - Analytics, Stats, AI, ML, etc.)

Each stage is therefore beneficial for the analytics business partner, but it is only within stage 4 where all of the questions can be answered and also where there is a mix of partnering skills such as collaboration, communication, and strategic leadership skills such as problem-solving and critical thinking.

Question 2: Tell me more about predictive analytics.

Answer: Predictive analytics is about utilizing mathematics such as statistics, predictions, and simulation techniques to identify the likelihood of future outcomes or the most likely outcome using an organization’s historical data. It's about drilling down in the data that is available to us by seeking out worthwhile opportunities and analyzing correlations and causations to solve business-related problems.

Honestly, for finance, it's more about a cultural shift from reporting the numbers to being a part of shaping the future. The focus is thus on how one can measure performance and help the business using quantitative analysis.

Managing predictive analytics (from a finance perspective) should also be easy enough for us to operate and should avoid complexity.

If you’ve ever sat in any A-Level or university math class, you’d recognize that predictive analytics builds on linear regressions, expected values, and probabilities. It can be used in finance in areas such as fraud detection, forecasting, and budgeting as well as mismatches within supply chains and operations.

Another issue worth noting is that predictive analytics can or cannot use AI and can be done through Excel (without using AI) as a starting point. Excel is still however not the most preferred tool to use because of its limitations and so applying analytics requires a step above Excel in order to truly leverage the benefits of AI and other analytics tools. Finally, this starting point should always be specified e.g., what am I trying to predict and why? And what will be the cost/benefit or impact in doing so?

Question 3: Are there new legal requirements that can be enforced that the algorithm knows nothing about right now?

Answer: Absolutely, AI analytics providers can update algorithms to include new statutory requirements such as legal updates, impending foreign trade regulations, and tax implications to provide updated predictions.

Question 4: Lastly what are some of the challenges of using analytics?

Answer: We know analytics is becoming increasingly important but executive buy-in could still be a major obstacle to progress any project. Also, data on the past doesn’t necessarily lead to the best decisions on the future even though AI is still in its infancy. Other challenges include confidence that analytics will make a sizeable impact depending on the size of the industry and the mindset change for finance teams from controlling to business partnering.

In closing, I’ll quote Robert Zwerling:

“AI/Analytics is a journey that is not hard, long, or expensive – it is a discipline.”

Surely the real test will no doubt come through implementation. Many thanks to Robert Zwerling, Steve Rosvold and Jesper Sorenson for being great lecturers.

As for crime forecasting using Machine Learning and the ethics involved in AI, this is beyond the scope of this article and will leave this discussion up to the experts!