When we want to build an ML model that's capable of playing Alpha GO. We're dealing with: Supervised Learning Unsupervised Learning Reinforcement Learning All of the above.
Suppose we want to build a model that predicts if a person has COVID-19 based on some specific features. For this reason, we go to different hospitals that give us the required data for people that tested positive and for people that tested negative. Which class of ML is best suited for this problem? Supervised Learning Unsupervised Learning Reinforcement Learning All of the above.
In the supervised learning process flow, we feed our data to an algorithm that produces a model.
Then to evaluate this model, we feed it the same data it learned from to see how much it was able to learn. True False.
After finding our cost function, our goal is to find the parameters that maximize it in order to make our predictions as accurate as possible. True False.
The main parameter set, that a linear regression model trains, must contain: The y-intercept The cost The coefficient and the y-intercept The gradient.
Gradient Descent is an optimisation method that finds the minimum of a function by calculating its derivative and setting it to zero to find its roots. True False.
We usually aim to increase the learning rate so that we learn faster. True False.
Logistic Regression is a regression method, as the name suggests. True False.
If the input of a sigmoid function is positive, the output should be mapped to 1, otherwise, it should be mapped to 0 (with 0.5 as a threshold between 0 and 1) True False.
This metric tells you how likely we are to be correct when we predict 1. Precision Specificity Recall F1 Score.
You want to build a model that predicts if a patient suffers from cancer or not. For this purpose, you collected data for several patients but you do not know which had cancer and which did not have it.
Which class of ML is best suited for this problem? Supervised Learning Unsupervised Learning Reinforcement Learning All of the above.
Which of the following statements is correct about parametric vs non-parametric models. Parametric models are much slower than non-parametric models. A parametric model is easier to be affected by outliers. A parametric model requires more data. Mostly, a parametric model's algorithm is easier to explain via common sense.
Predicting the price of the BITCOIN for the next day is a problem of: Regression Classification.
k-NN solves regression problems because it relies on the distance between 2 points True False.
We built a model to predict the temperature for the next 3 days. We obtained the following re [21.4, 25.4, 28.2]. However, the real temperatures were [20.5, 24.9, 29].
What is the MSE score of this model? 1.7 -0.6 0.56 0.73.
In Linear Regression, we perform Gradient Descent for each of the parameters a0 and a1. True False.
Since Gradient Descent is an iterative method, the ideal situation we usually stop when: We reach a max number of iterations We reach a max allocation of memory The values of the parameters start to saturate The values of the parameters become 0.
We use the mean squared error to evaluate the cost of a logistic regression model. True False.
This metric tells you how much we're predicting "1" when it's really 1. Precision Specificity Recall F1 Score.
Recall is the same as TPR. True False.
|
