Understanding How Machine Learning Works [without math]

Understanding How Machine Learning Works [without math]

Most executives I talk to still think of machine learning as some kind of robot-ruled world like Skynet out of the Terminator movies.

While machine learning can certainly get advanced, it is quite the opposite of the antagonistic artificial intelligence (AI) network that seeks to destroy the world in the Schwarzenegger series.

Andrew Ng, a foremost mind in the AI field, explains artificial intelligence well when he states that “some input data (A) is used to quickly generate some simple response (B).”

It is important to understand that artificial intelligence and machine learning are two different things, although the two terms are often used interchangeably.

AI is when we program machines to carry out tasks in ways we consider intelligent. Machine learning is an application of AI, but is more broadly based on the idea that we should give computers access to data and let them learn on their own.

In the software sense, these terms frequently come up when the topic of Big Data is discussed. Let’s look at some examples of machine learning and look into how it works.

What are some Examples of Machine Learning?

You are probably using machine learning to your advantage everyday. Take Netflix for example. Once you watch a few different types of shows the system uses machine learning to recommend additional shows you might like based on your watch history.

Spam detection within your email is also a simple example of machine learning. A more advanced example of smart machines learning about their surroundings is autonomous vehicles.


Here are some additional examples of machine learning and the response chain:

Source: Harvard Business Review

How to Identify Machine Learning Opportunities in Your Business

Machine learning is used to simplify our lives or make them more productive. The same is true with machine learning in professional organizations, and is the reason software, often masked with the buzzword Big Data Analytics, is created.

Here are the steps you can take to imagine a machine learning scenario in your business:

1

Identify tasks you perform regularly at work

Most people, and companies, have tasks that they carry out regularly at work. Take time tracking for example; this is a task many people at an organization carry out every day. Time tracking can also involve project management, billable hours, how the workforce is optimized, and determine who may be available for a task or who is assigned to a future task.

2

Identify other people in your organization who perform the same tasks

Ask yourself the following: if 10 of your co-workers perform the same task, would they all agree on the expected outcome? If humans can’t agree on an outcome, machines can’t be programed to use algorithms to learn. If you can agree, move on to the next step.

3

Determine if your organization has documented the completion of this task

If your organization has been successful at documenting the task, you are armed with some data that a data scientist can input into an algorithm. This would be your initial training data set. If not, you can start collecting data of the task based on the above criteria.

To continue with the time tracking example, you could compile information on billable hours versus non-billable hours, compare project estimates to actual time billed, or evaluate historic data on how many hours a particular team of employees has been working on tasks.

4

Work with a data scientist to determine if tasks can be automated

Once you’ve established that you have some training data sets to work with, consult with a data scientist to go over the task and see if machine learning might help with any automation of data.

For time tracking, your data scientist team may be able to use your time tracking data to help calculate project burndown rates based on historic trends of employees or project teams.

5

If automatable, can the task provide more value to your business or customers?

Now that you are armed with some data and a data scientist has reviewed it, is automation an option? If so, would it benefit your customers or business? What is the worst-case scenario if this task were automated? Finally, if the machine learning algorithms are only 65% or 70% accurate, what effect does this happen on the business? In other words, what is your accuracy threshold?

Machine Learning using Time Tracking and Business Intelligence

I used the above time tracking example above because that is exactly what Allocable was built to provide businesses.

Allocable understands patterns in time tracking data -- either imported from other software or inputted into our software by its users -- to help provide insight into forecasting trends for multiple areas of your business.

Here are some ways you can use machine learning and time tracking around some key performance indicators:

WAYS TO USE MACHINE LEARNING & TIME TRACKING AROUND KPIS


Budgeting: after assigning billable hours to employees, you can utilize machine learning software to see if projects are profitable and understand real-time burndown metrics. Allocable does this visually through a dashboard.

Planning: billable or nonbillable, emails, research, meetings, etc. Wouldn’t it be nice to see what your workforce is spending its time on each day, week, month, or lifetime of a project?

Forecasting: with historic data on burndown rates you can forecast the outcome of future projects with more confidence. This allows you to save time, hire more, sell more, and make better decisions.

Optimization: seeing how your workforce spends its time day-to-day will allow you to monitor productivity and optimize your workforce so they are working at their best.




Imagine having one place to track your time and expenses, handle resource allocation, view project and workforce analytics all with beautiful BI dashboards built for the different roles in your business. Allocable provides a complete visualization of your workforce and project productivity data. We empower you to turn information into actionable insight to optimize performance with certainty.

View a demo of Allocable to get started today.

About

Allocable is a cloud-based automated time tracking and business intelligence (BI) software platform that provides  a complete visualization of your workforce and project productivity data empowering you to turn information into actionable insight to optimize and forecast performance with more certainty.

Headquarters

Allocable
319 Clematis St. Suite, #508
West Palm Beach, FL 33401
social@allocable.com
+650-503-3577

Follow Us

  •      
  •