Machine Learning Specialization Training

  • Learn via: Classroom
  • Duration: 3 Days
  • Price: Please contact for booking options
We can host this training at your preferred location. Contact us!

It is impossible to say "This model will work best for this kind of data". If we can say that, we would not need people to try and build models, we could automate the process. So how can we choose optimal model for our problem and how can we make that process faster ? How can we make our training so that when we ship our model and it is live, it will be generalizable ?

After looking at those subjects we will continue with using our trained model to understand our data more and then build a stronger model with those insights and then use a new model to understand our data more. After iterating over this process couple of times we will have strong insight into our data and we will have a better model. Also, we will see which prediction of our model we should trust more and will investigate powerful tool that allows us to go beyond prediction and make simulation using our model and answer how changing our input will affect our outputs (this is very valuable in cases like you want to have more customer coming to your business and want to reduce the number of customers leaving etc...)

This course will give you the intuitive understanding of the concepts without memorization. This course combines both theory and practice to give students deep understanding of the subject and hands-on experience via coding.

Students must have experience in coding in Python language and are expected to have knowledge about Pandas and Numpy Libraries. If they don't, taking python coding and python for data science lectures before this course is recommended.

Individuals comfortable with basic programming and machine learning looking to master machine-learning techniques. 

Practical Machine Learning 

  • What is machine learning ?
  • Regression and Classification
  • Regression and Classification using Decision Trees
  • What is the cons of using Decision Trees ?
  • Understanding Bagging
  • Ensemble methods - Random errors
  • Understanding Random Forests
  • Why do we need train-validation-test set instead of just train set?
  • No. Cross validation and random sampling are not always a good idea ?
  • Careful selection of Validation set
  • What is R^2 and where to use this ?
  • Logic behind Baseline Model
  • Do not use all of your data while hyperparameter tuning. How to take subsample, and how to understand it is representative ?
  • Profiling - How to detect which parts of your code are slow

Model Driven EDA

  • Model is not just something that just makes predictions - What is model Driven EDA.
  • Prediction confidence using standard deviations
  • Feature importance while keeping interactions between features active -
  • Permutation Feature Importance
  • Okay, this is our prediction, but how can we change the outcome?
  • Simulating different scenarios using Partial Dependence
  • Extrapolation problem
  • How to reduce extrapolation problem ?
  • How can we understand we split validation set right ?

Contact us for more detail about our trainings and for all other enquiries!

Upcoming Trainings

Join our public courses in our Istanbul, London and Ankara facilities. Private class trainings will be organized at the location of your preference, according to your schedule.

Classroom / Virtual Classroom
27 August 2024
Istanbul, Ankara, London
3 Days
By using this website you agree to let us use cookies. For further information about our use of cookies, check out our Cookie Policy.