How To Use Machine Learning For Your Small Business | by Devansh – Machine Learning Simplified | March 2022

Recently, a reader + viewer of my content contacted me about machine learning and some specific concerns/challenges he was having with his machine learning journey. He asked me a list of questions, which I gladly answered (he asked good, well-structured questions). Among the questions was a particular video request:

The answer to question 2, if you were wondering, is absolutely. Shocking I know

I realized this would definitely be an interesting topic for me to cover. I’ve covered the technical details of Machine Learning, but I’m not talking too much about the commercial implementation side. Especially for small organizations, which have a much lower margin for error/experimentation than larger companies. In this article, I will explain how you can implement machine learning in your small organization. We’ll cover the process, end to end, going over some of the considerations and challenges you’ll face along the way.

“Hi Devansh, I just came across your article and your youtube channel, I am also learning machine learning, I want to connect with you as I believe you are the right person to guide me on my machine learning journey “

— The exact message he sent me

As someone who’s been involved in small organizations, huge multinationals, and everything in between (including being involved in a growing startup), I’m going to give you the perspective from multiple angles. If you have experience in this area, be sure to share it in the comments below. For this article, I will focus more on the process than the technical details, as the technical steps are context dependent and can be learned on the internet. This includes a certain machine learning blog, a YouTube channel, and a daily coding newsletter. There aren’t many good resources that talk about the process.

For the purposes of this article, I will use the example of an online edutech/coaching service. This is just for convenience, and the principles discussed are universal.

It sounds trivial, but it really isn’t. When defining your problems, it is important to specify several details. Where would you get possible data sources? Is there a set of data the client already has that they want to analyze, or will you need to figure out the features? The first is simpler (I’ll be making a video about my favorite pipeline soon). For the latter option, you will need to do a lot of research on the domain. Suppose you want to identify customers who are likely to be unsubscribed based on their behavior. Look at the features different industries have used. Talk to a few domain experts in your own industry to see what features might be relevant and how you can relate the features to the edutech domain.

Understand why potential customers interact with your products. This will optimize your approach.

Another important issue concerns finances. What is your profitability plan? For example, you can create a company that manages the end-to-end process, from data collection to recommendations. It takes a bigger investment but will give you a ton of experience in the field. It will also allow you to approach organizations and show them exactly how you can help them. For such a setup, you will have to organize an initial amount of money as a track until you can achieve profitability. This money can come from loans, investors or personal savings. Each of these methods has pros and cons, so be sure to review each and decide which is best for your situation.

Your problem/area discovery will help you find the solutions. Photo by Sigmund on Unsplash

When I usually worked with smaller organizations, I asked the organization to manage the data collection. I would do the domain research, tell them what to collect, and then work with that data. This approach saves you a lot of headaches. This approach worked for me because I was using these small projects to build my experience and do bigger projects. Either approach will work just fine for you.

One of the most important aspects you want to cover when implementing for small organizations is setting minimum baselines. Chances are your project will run out of funding/resources before you can try out all of your ideas. You want to schedule this event. What are the minimum acceptable results? What would compensation look like? Other Provisions. These conversations can be awkward, but need to happen before a lot of time, energy, and resources are poured into the project.

The next step is to create your machine learning pipeline. Creating a strong pipeline is very important as it will help you integrate different imputation policies, model training protocols, and other sources of variance.

Watch this short video to understand machine learning pipelines: https://www.youtube.com/watch?v=3FoJTBjEb-U&ab_channel=Devansh%3AMachineLearningMadeSimple

When developing your pipeline, be sure to use a very truncated version of your dataset. This will save you a lot of time when testing your pipeline. The objective at this stage is NOT to do the analysis. It’s just to make sure your pipeline is working. Watch this video for a project that will give you the skills to build these pipelines. I will be making a video covering the details of such a pipeline soon, so be sure to subscribe

Next comes the unsexy part of this process. You will have to do a lot of trial and error. Once you have run your pipeline on your complete datasets, you’ll have a ton of reports to sift through. Understand different data imputation policies and other moving parts and assess their impact on datasets. You will discover the strangest things.

One of the things you will notice is that you will have to reset many features. You will have to test and drop a large number of them. Each time your dataset changes, you will need to rerun the pipeline. This is one of the reasons I recommend people use the smaller models. In this constant process of rebuilding and iterations, you probably won’t be able to afford the expensive models used.

Trial and Error will also give you insight into the code and processes allowing you to build and extend your solutions to fine-tune the specifics.

Once your theoretical solution works, it’s time to deploy it in practice. You will often face challenges integrating data sources into pipelines to fully automate the entire process. Depending on the domain and the nature of the features used, you may need to configure feature recycling and monitoring protocols.

Of course, setting up machine learning solutions for small organizations can be tricky. They often lack a large number of resources that allow for larger operations, memories, and data analysis. Luckily, you can mitigate this by spending a lot of time in step 1, really breaking down the components.

A cool thing is that a good personal project is quite similar to this article in scale and operation. This is why I recommend absolute beginners to use real data to build their machine learning projects. Experience in thoroughly analyzing the domain, identifying features and procedures, and building all elements of the ML pipeline will be crucial.

To really get good at Machine Learning, a foundation in software engineering will be crucial. They will help you conceptualize, build and optimize your ML. My daily newsletter, Coding Interviews Made Simple, covers topics related to algorithm design, math, recent tech happenings, software engineering, and more. You can view the program here

I created Coding Interviews Made Simple using new techniques discovered through tutoring several people at high tech companies. The newsletter is designed to help you succeed, saving you hours wasted on the Leetcode grind.

The newsletter has helped a ton of people succeed in tech

You can check out my articles and reviews of other successful students for more success stories and detailed tips for success

To help me write better articles and understand you, complete this (anonymous) survey. It will take 3 minutes maximum and will allow me to improve the quality of my work.

Do not hesitate to contact me if you also have interesting jobs/projects/ideas for me. Always happy to hear from you.

For monetary support of my work, here is my Venmo and Paypal. Any amount is appreciated and helps a lot. Donations unlock exclusive content such as paper scans, special codes, consultations and specific coaching:

Venmo: https://account.venmo.com/u/FNU-Devansh

Paypal: paypal.me/ISeeThings

Use the links below to check out my other content, learn more about tutoring, or just to say hello. Also check out the free Robinhood referral link. We both get free stock (you don’t have to put any money down), and there’s no risk to you. So not using it is just wasting free money.

Check out my other articles on Medium. : https://rb.gy/zn1aiu

My YouTube: https://rb.gy/88iwdd

Contact me on LinkedIn. Let’s connect: https://rb.gy/m5ok2y

My Instagram: https://rb.gy/gmvuy9

My Twitter: https://twitter.com/Machine01776819

If you are preparing for coding/technical interviews: https://codinginterviewsmadesimple.substack.com/

Get free stock on Robinhood: https://join.robinhood.com/fnud75

Sherry J. Basler