What Is Holding You Back From Your Machine Learning Goals?

Identify and Tackle Your Self-Limiting Beliefs and
Finally Make Progress

I get a lot of email from developers and students looking to get started in machine learning.

The first question I ask them is what is stopping them from getting started?

I try to get to the heart of what they are struggling with, and almost always it is a self-limiting belief that has halted their progress.

In this post, I want to touch on some self-limiting beliefs I see crop up in my email exchanges and discussions with coaching students.

Maybe you will see yourself in one or more of these beliefs. If so, I urge you to challenge your assumptions.

Don't give up

Don’t Give Up
Photo by brendan-c, some rights reserved

Self-Limiting Belief

A self-limiting belief is something that you assume to be true that is limiting your progress. You presuppose something about yourself or about the thing you want to achieve. The problem is you hold that belief to be true and you don’t question it.

Steve Pavlina lists 3 types of self-limiting beliefs in is post: Dissolving Limiting Beliefs:

  • If-then Beliefs: e.g. If I get started in machine learning, I will fail because I am not good enough.
  • Universal Beliefs: e.g. All Data Scientists have a Ph.D. and are mathematics rock gods.
  • Personal and Self-Esteem Beliefs: e.g. I’m not good enough to be a machine learner.

You’re probably a logical and rational thinker. Apply those skills to your own beliefs about your goals and aspirations in machine learning and challenge them.

Waiting To Get Started

I think the biggest class of limiting belief I see is the belief that you cannot get started until you have some specific prior knowledge. The problem is that the prior knowledge you think you need is either not required or is so vast in scope that even experts in that subject don’t know it all.

For example: “I need to KNOW statistics“. See how ambiguous that belief is. How much statistics, what areas of statistics and why do you need to know them before you can start your investigation into machine learning?

Below are some of the more common self-limiting beliefs of skills or prior knowledge that must be obtained before you can get started in machine learning.

I can’t get into machine learning until…

  • …I get a degree or higher degree
  • …I complete a course
  • …I am good at linear algebra
  • …I know statistics and probability theory
  • …I have mastered the R programming language

You can get started in machine learning today, right now. Run your first classifier in 5 minutes. You’re in. Now, start blocking out what it is from machine learning that you really want?

I have written about some of these before, for example:

Awaiting Perfect Conditions

Another class of self-limiting belief is where you are waiting for the perfect environment or conditions before taking the leap. Things will never be perfect, leap and make a mess, then leap again.

I can’t get started in machine learning because…

  • …I don’t have the time right now
  • …I don’t have a fast CPU, GPU or a bazillion MB of RAM
  • …I am just a student right now
  • …I am not a good programmer at the moment
  • …I am very busy at work right now

It does take a lot of time and effort to get good at machine learning, but not all at once and not all at the beginning.

You can make good progress with a few hours a week, or tens of minutes per day. There are plenty of small snack-sized tasks you could take on to get started in machine learning. You can get started, it is just going to take some sacrifice, like all good things in life.

Struggling or Tried and Failed

The third class of limiting belief is that where you have made a small start but you are struggling or have failed to achieve your goal.

This is a tough one. Machine learning is hard but no harder than other technical skills like programming. It takes persistence and dedication. It’s applied and empirical and demands trial and error.

I can’t get into machine learning because…

  • …I feel overwhelmed
  • …I don’t understand x
  • …I will never be as good as y
  • …I don’t know what to do next
  • …I can’t get my program to work

My advice is to cut scope or change direction. I advocate small projects as often as I can because the methodology has been so successful for me.

What is your self-limiting belief?

Do you have a self-limiting belief? Think about it. What are your goals and why do you think you are not there yet?

Do you have a goal to get into machine learning, to become a data scientist or a machine learning engineer but have not taken the first step?

  • Are you waiting to acquire some perfect set of skills before getting started?
  • Are you waiting for the perfect conditions before getting started?
  • Have you taken a first step and abandoned the trail?

Where do you want to be and what are you struggling with?

68 Responses to What Is Holding You Back From Your Machine Learning Goals?

  1. Aman Tandon November 12, 2015 at 1:41 pm #

    Exactly, we should go beyond our limits and we should think there is nothing which we can’t do. I have weak aptitude but have very good ideas to implement various things using machine learning.

    And I really appreciate your work. Thank you Jason

  2. Roberto May 18, 2016 at 8:31 am #

    This page is pure awesomeness

  3. Rex May 31, 2016 at 11:00 am #

    Love what you wrote.
    I just forwarded this article to my friend who has been coming up with all sorts of excuses to delay his learning of Python.
    I hope this could slap him so hard in the face, that he won’t be able to look at himself in the mirror for the rest of his life before he starts moving again.

  4. Kleyn Guerreiro June 7, 2016 at 5:01 am #

    Beeing a data journalist would be my preferred approach to encourage people to start with ML and this is what I couch people who watch my speeches…After a few projects learning how to find data, build visualizations e tell stories, then algoritms and ML would be the next step…

    • Jason Brownlee June 14, 2016 at 8:23 am #

      Great suggestion Kleyn, although it does presuppose skills in writing.

  5. Bilal Malik June 20, 2016 at 8:05 pm #

    I knew nothing about Machine Learning till 2010 till I donwloaded PLS toobox and boom got the results for my data using SVM.(At this stage I didn’t know much about SVM and may not much about linear regression itself. But I manged to use SVM for determination of glucose concentration and was able to publish few papers in journals on chemometrics.

    But till date I have not been able to build on my learning i.e. do something about ML and get published in this area.
    I always think it may be my math skills. although I tried to learn linear algebra from Andrew Nag’s Course but …………… and did a statistics course from Coursera and yes started working using Matlab ….
    But I feel I am not able to go much beyond this.

  6. Maanav July 1, 2016 at 12:37 am #

    I am a 9th grade student from India, am deeply interested in machine learning. I have mastered the fields of linear algebra, probability and statistics, but I wanted to inquire whether it is the right age for me to pursue ML. Should I wait a few more years? Or can I begin right now?

    • Jason Brownlee July 1, 2016 at 5:41 am #

      If the field interests you then why not get started.

    • S Kotrappa February 26, 2017 at 1:41 am #

      Congratulations !! Maanav Go ahead there is no age for learning skills & you can become master and expert by the time you come to UG , all the best!!

  7. Jessica September 4, 2016 at 7:27 am #

    Thank you for sharing! I am a graduate student majoring Information Systems. Most of time I have to use machine learning techniques to solve business problem. Since I focus on academic and research area, more advanced and novel methods should be created by myself. And I feel that is very difficult for me. For now, I am researching on Bayesian Network and calculate parameters by using EM algorithm to predict the label. Sadly, the results are not quite good compared with HMM model. Still struggling on my work. But your website really helps me a lot. Hope you could publish some more interesting machine learning techniques and articles especially on how to improve the original algorithm! Thank you very much!

    • Jason Brownlee September 4, 2016 at 8:06 am #

      I’m glad you’re finding it useful Jessica, thanks.

  8. Baouche rafik September 28, 2016 at 9:11 pm #

    Machine Learning in geophysical is a chalenge for predictions of Parameters which approach the model in earth sciences. Using the well log data and process the rules, the prediction of reserves evaluation in oil field can be possible.

  9. Amrit October 2, 2016 at 5:59 pm #

    This is really helpful to start for a beginner .
    Encouraging and well defined.
    Thank you.

    • Jason Brownlee October 3, 2016 at 5:18 am #

      I’m glad to hear that you found it useful Amrit.

  10. Sagar October 6, 2016 at 11:10 pm #

    This is a helpful post and motivational. I am going for PG in Data Analytics and this post motivated me to overcome many self beliefs and doubts. I was in double minds on whether to go for PG or not considering lack of practice in Maths/Stats and new concepts of ML (R and Python).
    But agree to your writing that we can overcome these fears and start learning.
    Thanks for this post 🙂

  11. Saad Yaseen October 9, 2016 at 6:31 pm #

    I am not so good with Mathematics. I’ve always scored C grades in most university math courses. However, I am quite excited about ML, I am adept at Java programming. Do I have to have better grades in Mathematics in order to learn ML?

  12. Mark Kaghazgarian November 17, 2016 at 9:16 pm #

    This article made my day and gave me enough motivation to keep going ahead in learning machine learning since I was overwhelmed by many resources out there which usually asked for prior knowledge in Mathematics and statistics

  13. Diego November 30, 2016 at 5:00 am #

    Thanks for inspiring me. Greetings from Ecuador

  14. Krishna November 30, 2016 at 8:17 pm #

    Great post !! its always interesting to understand our own mind play and overcome the limits with the same will power 🙂 And this applies to anything new to start with.

  15. Yashraj November 30, 2016 at 11:31 pm #

    Though this article is written by keeping machine learning in mind, this article can be co-related any new thing to learn which we think is gigantic effort and out of reach for us.
    Thanks Jason for such inspiring thoughts

    • Jason Brownlee December 1, 2016 at 7:29 am #

      Thanks Yashraj, I’m gald you found it useful.

      I think it is such an import area that few are talking about.

      It all comes down to mindset, the tools are already there and ready to use.

  16. Taye December 2, 2016 at 9:07 pm #

    Thank you Jason for such a wonderful inspiring thoughts. I hope any one can even use these principles in his every day life other than Machine Learning.

  17. Arnold January 5, 2017 at 1:54 am #

    thanx Jason, am one of them i tryed a small project with no success……ended up giving up but will give it a try again

  18. Imane January 14, 2017 at 4:18 am #

    Thanks for advices, I’m now ready to get start my first leap

  19. Murali February 13, 2017 at 5:26 pm #

    Till now i was very confusion and i had the notion that machine learning is not for me because i don’t know math stuff with me. This post inspired me so much, made me to come out of my inner fear. Thank you very much Jason.

  20. S Kotrappa February 26, 2017 at 1:46 am #

    Wonderful Jason for your service and resources , motivation you provide us in Machine Learning, thanks

  21. Abhishek sharma March 3, 2017 at 6:20 am #

    Thanks for making me fearless about machine learning through this amazing inspirational content and you are providing such a phenomenal resources on machine learning for a beginner as well as for expert.
    Thank You So much, Mr Jason Brownlee

    • Jason Brownlee March 3, 2017 at 7:46 am #

      You’re welcome Abhishek.

      Thank you in return for your support! Without readers like you I’d be out here talking to myself.

  22. Gautham March 4, 2017 at 11:51 pm #

    This is the best getting started page I’ve ever seen, In all my self learning journey.

  23. Jijo March 11, 2017 at 4:05 am #

    I not a programmer,but i want to learn Machine Learning.Should i first master Python or other languages to start ML

  24. Uche April 3, 2017 at 8:51 pm #

    How do I get started.

  25. Uche April 3, 2017 at 8:52 pm #

    I want to start with R. How do I get started

  26. Abhishek Kumar April 8, 2017 at 3:58 pm #

    Can you be my mentor please? I can do anything to learn from you. I know this website helps a lot in machine learning already, but I would like to go beyond this and to achieve extreme level in my life. Please mentor me even though it takes anything.

    • Jason Brownlee April 9, 2017 at 2:57 pm #

      You can learn from my tutorials, blog posts and books. It is a way I can help many students at once.

  27. T.D.Nuwan Chathuranga April 21, 2017 at 4:22 am #

    Hi Jason,

    Thank you for your good thoughts and influence for a become a good ML Engineer,

    Learning ML in good = 1/Self Limiting Belief

    Thank you again

  28. Sathish Kumar C May 3, 2017 at 2:30 pm #

    Hi Jason,

    Thanks for your motivational thoughts,

    To become a datascientist, I feel like need to be technically(I am little strong) and analytically (I am not week, need to explore more analytical equations) strong.

    Learning ML is good and very intersting..

  29. Bharat Poptwani July 26, 2017 at 8:49 pm #

    Hi Jason,

    You have put me in a dilemma haha. I am a software developer at India’s Biggest ticketing platform, BookMyShow. I have good enough challenges on my plate in both client-side (We are moving our Native apps to React Native) as well as server-side. But I had this sudden desire for uninterrupted learning, I do keep learning new developments in web development world, but the thing is I am restricted to the time that I find after working hours and on weekends. So I was thinking of doing MS in US specifically in Machine Learning.

    Even though I will take a huge educational loan, my reasons for doing MS in US were:
    1. First and foremost I will be learning uninterrupted for 2 whole years.
    2. Highly qualified faculties.
    3. Access to the best resources available (The myth that you busted, internet is the best resource available, thanks for reaffirming that. Even I believe so because I am a self-taught programmer myself)
    4. Spending time with other similarly interested and enthusiastic individuals.

    Now, I would like your opinion on if I should make that trip and live two tough years to devote myself to machine learning? I see points 1,2 and 4 as solid reasons, what do you think?

    • Jason Brownlee July 27, 2017 at 8:03 am #

      This is your choice and you must use your freedom to make it.

      I wish you the best of luck and would love to hear what you decide.

  30. Satya Prakash Sharma July 30, 2017 at 12:46 am #

    I have a Burning Desire to learn a machine Learning , Already
    I am Good in Android Developing , Already I developed an Android App , that is “SafeWallet One 4 All Password Manage”.

    Jason I like your blog , really this is helpful. can you guide me .

  31. FIKIR August 28, 2017 at 4:04 pm #

    hello im masters sudent I took ML course but still im confused how build the model how selct algorithims how apply …soon Till now i was very confusion i have low confidence low low self-esteem and most of the time i feel I don’t have the time right now because im student.now i want do my research around ML im interesting when im searching different sites i get ur site im so excited i hope this post inspired me so much,highl improve my confidence not only by ML by all thing and gave me enough motivation. Thank you very much Jason!!!!!!!!!!!

  32. Renato Caetano September 22, 2017 at 8:23 pm #

    You help me a lot… Thank you Jason Brownlee!

  33. Aadrsh September 28, 2017 at 4:53 pm #

    That’s so motivative.
    Thank you, Sir.

  34. Aniket October 1, 2017 at 3:25 am #

    Hi Jason,

    I have a question about how to get started in machine learning. The thing ehich frustrates me and even hoding me back till today is that there are thousands of courses out there online having different contents(topics) to go through and I don’t understand totally which course I should go through to understand the nuts and bolts of machine learning. Even quora answers about “how to get started” don’t help me to take a good course.
    I have also taken up machine learning couse on coursera by Andrew Ng but it covered some aspects of machine learning.
    So, please help me to guide a good course in machine learning.

    • Jason Brownlee October 1, 2017 at 9:08 am #

      Great question.

      There are many ways to get started, you need to find the one way that works best for your preferred learning style.

      I teach a top-down and results-first approach to machine learning.

      My best advice is broken down into a 5-step process, summarized here:

      I hope this helps as a start.
      I’m here to help as you have more questions.

Leave a Reply