Math bot does some specific calculations in Python, so that you can do other things :)
Math bot will ask you what you wish to calculate, and can remember previously given arrays and data points to work with again. Its memory is reset every time you finish working with Math bot.
Although capitalisation doesn't matter when requesting a specific calculation, spelling and any extra spaces will confuse math bot. Below are a list of calculations Math bot can do, and what you need to type to trigger that calculation.
To finish, type "end".
- For this, type "accuracy".
Math bot will ask for the number of correctly classified data points and then the total number of data points and return the accuracy.
- For this, type "precision".
Math bot will ask for the number of true and false positives and return the precision.
- For this, type "recall".
Math bot will ask for the number of true and false negatives and return the recall.
- For this, type "mse".
Math bot will calculate the MSE after asking for a list of x's and corresponding y's
- For this, type "mae".
Math bot will calculate the MAE after asking for a list of x's and corresponding y's
- For this, type "f1".
Math bot will calculate the f1 given the number of true positives, false negatives and false positives. Math bot does not remember any previously given false/true positives/negatives.
- For this, type "knn weights".
Using x's and corresponding y's, Math Bot calculates the Euclidean distance to calculate the K Nearest Neighbour weights.
- For this, type "sd".
Math bot will ask for a list of values and will calculate the SD.
- For this, type "mean".
Math bot will ask for a list of values and will calculate the mean.
- For this, type "euclidean distance" or "euclidean".
Math bot will ask for a list of xs and ys and will calculate the Euclidean distance.
- For this, type "manhattan distance" or "manhattan".
Math bot will ask for a list of xs and ys and will calculate the manhattan distance.
- For this, type "gaussian probability" or "density".
Math bot will find the probability density at a particular point given a list of numbers
- For this, type "gaussian mixure model" or "mixture".
Math bot estimates parameters of a Gaussian mixture model probability distribution when you give it some data. It will plot all the data in blue and then the means in pink on a scatter graph. Currently only supports 2 arrays (xs and ys for example) Please read this documentation for more information: https://scikit-learn.org/stable/modules/mixture.html#gmm
- For this, type "bayes".
It uses Bayes' theorem to allow you to calculate the prosterior, likelihood, prior, marginal and the joint probability.
You will need to know 3 of the 4 parts of the equation, but Math bot will remove the hassle of rearranging this for you. Mathbot will ask you which of the posterior, marginal prior or likelihood you want.
IF YOU CAN REARRANGE BAYES THEOREM, IT IS HIGHLY RECOMMENDED THAT YOU CHOOSE THIS OPTION OVER THE JOINT PROBABILITY EQUATION.
Mathbot will ask you if you want to calculate a conditional probability. This means something like A|B or A|B,C
If you choose NOT to calculate a conditional probability, mathbot will ask you the probability you want to calculate e.g. A, and then "what it relys on".
This is dependent on the Baysian network, like the following:
So if you wanted to calculate P(S) you would type 'S' as what you wanted to calculate and then 'C' for what S relies upon.
If you choose TO calculate a conditional probability, you must tell mathbot the full joint equation. It will do any marginalisation possible. This equation can be derived from the Bayesian network that you should have been given. So, for example, the joint equation in this example image's case is P(C)P(S|C)P(R|C)P(W|S,R), because C has nothing going into it, S depends on C because of the direction of the arrow, which is same as R. And finally, W depends on both S and R.
It then steps through all of the probabilities you should also know (it will likely give you redundant information sometimes, you can ignore this) and then provide you with the solution.
- for this, type "point crossover"
Works with binary and letters. Mathbot asks you for the parents and the number of cross-over points, and then the cross-over points themselves. Then it just does the point cross-over. If you have decimal numbers, you would need to convert them to binary.
- for this, type "swap mutation"
Works with binary and letters. Mathbot asks you for the parents and the number of swaps you want to make (this MUST be an even number), and then the indexes of the swap(s) themselves. Then it just does the swap mutation. If you have decimal numbers, you would need to convert them to binary.
- for binary to decimal, type "binary conversion". For decimal to binary, type "decimal conversion"
Converts from binary to decimal and vice versa.
- for this, type "fitness"
Mathbot will ask you for a fitness function and some values, which you must enter in the order they are on the roulette wheel. It will then rescale the values if required. Mathbot will then output the:
- "raw fitness" for the results that come straight from the fitness function,
- "rescaled" for any rescaled values (mathbot will also tell you if this step is not required), and finally
- "relative fitness" for the relative fitness.
IMPORTANT NOTE: The fitness function must be typed with the following rules:
No spaces
The equation must be written in terms of x e.g. x**2 is okay but y**2 is not.
It cannot handle logs.
Sometimes something strange can happen where if you copy and paste the equation from somewhere else,
Mathbot has some kind of breakdown. I'm not really sure how to prevent it so I recommend typing the
equation out yourself with the following syntax:
* = multuply
\ = divide
** = power
+ = add
- = subtract or negative
i.e. $ −0.3x^3 + 2x^2 − x + 12 $ is written as −0.3*x**3+2*x**2−x+12
- for this, type "prior"
Calculates the prior probability of a class of data points in a data set
- for this, type "linear"
This gives you the gradient, y-intercept, SSE (Sum Squared Error) and R Squared using xs and ys that Mathbot will ask you for. It also provides a graph as well. It uses some old code I wrote myself some years ago. I think it works anyway.
- for this, type "naive"
Math bot will ask you what you do or don't know and then calculates which class a new data point will be in, using the log-likelihood. Math bot will tell you the log-likelihood for each class, and the prior probability if you want.
IMPORTANT: Math bot will automatically 0 index (so start at 0) with the classes, which may not be the same as your notation. When it tells you the selected class, you will need to keep this in mind. For example, if you have 3 classes, 1-3, mathbot will number them 0-3 and therefore "class 2" is actually a different class in this example.
- for this, type "clustering"
Clusters data points via K-Means per iteration and provides new centroids. If you want to know convergence, increase the iteration count and see whan the value stops changing for the centroids. Typing 0 when prompted for how many initialised centroids will show you the optimised number of centroids. It may give a warning, you can ignore this.
- for this type "value" or "value iteration"
Can do value iteration in a 2D and 1D grid worlds. Cannot move in diagonals. Can move in only up-down, left-right or both. It will ask you for terminal states which includes goal states.
IMPORTANT NOTE: Mathbot indexes Grid World in the following way: [row, column] and 0 indexed.
This may be different to the way it is indexed in the question. BE WARY.
