Vectorkit is part of a series of python implementations of mathematical concepts of AI from the scratch. It is a pet project. Even though it gets packaged to pypi, it is a work in progress, will undergo a lot of changes, and not (conciously) optimized for large scale computation. You can play with its javascript counterpart in the browser at https://vectorkitweb.netlify.com/.
components
- a list of the components of a vector
dimensions
- the dimension of the vector, or count of its components
min
- the minimum component
max
- the maximum component
sum
- the sum of the components of a vector
memsize
- the size of a vector in memory
add
- Adds two vectors
append
- Appends new components to a vector
concat
- Merges two vectors into a single vector
corr
- Returns the correlation of two vectors
cosinesim
- Returns the cosine similarity between two vectors
cost
- Computes the squared error cost function, under the assumption that one vector is the ground truth and the other is the compared.
cov
- Returns the covariance between two vectors
crossmul
- Returns the cross product of two 3-dimensional vectors
describe
- Returns a description of a vector, including its dimensions and memory size
distance
- Returns the euclidean distance between two vectors
dotmul
- Returns the dot product between two vectors
ediv
- Returns the element-wise quotient of two vectors
emul
- Returns the element-wise product of two vectors
insert
- Inserts a new component at a specified index
jaccard
- Returns the jaccard similarity between two vectors
leakyrelu
- Passes vector through the leaky version of Rectified Linear Unit
leastdev
- Returns the Least Absolute Deviations(L1 Norm) between to vectors
leastsq
- Returns the Least Squares(L2 Norm) of two vectors
mae
- Returns the mean absolute error between two vectors
mbe
- Returns the Mean Bias Error between two vectors
magnitude
- Returns the magnitude of a vector
mean
- Returns the mean of the components of a vector
minmax
- Returns a variant of a vector which has been normalized using standard min-max feature scaling
minmaxmean
- Returns a variant of a vector which has been normalized using standard mean and min-max feature scaling
mse
- Returns the mean square error of two vectors
pad
- Appends zeroes to vectors to a specified length, in-place
padded
- Returns a new vector with zero appended to it to a specified length
pararelu
- Passes vector through the parametric version of Rectified Linear Unit
pop
- Removes a component at a specified location
relu
- Passes a vector through a Rectified Linear Unit function and returns a new vector
reverse
- Reverses the direction of a vector in-place
reversed
- Returns a variant of a vector with reversed direction
rmse
- Returns the root mean square error between two vectors
rsquare
- Calculates the R square error between two vectors
sdiv
- Returns a new vector, which is the quotient from a scalar division of a vector
shuffle
- Shuffles vector components in place
shuffled
- Returns a new vector with shuffled version of a vector's components
sigmoid
- Passes a vector through a logistic sigmoid function and returns a new vector
softmax
- Passes a vector through a softmax function and returns a new vector
smul
- Returns a new vector, which is the product from a scalar multiplication of a vector
std
- Returns the standard deviation of the components of a vector
stdnorm
- Returns a variant of a vector which has been normalized using the z-score
subtract
- Returns a new vector, which is the result of the subtraction of one vector from another
subvec
- Returns a new vector which is a slice from the original vector
tanh
- Passes a vector through a TanH function and returns a new vector
to_list
- Returns a list of the components of a vector
to_tuple
- Returns a tuple of the components of a vector
unitvec
- Returns a new vector which has been scaled to unit length
vector_eq
- Returns the vector equation of a line between two vectors
isovector
- Returns a vector of a specified length containing the same component throughout
randvec
- Generates a random vector of specified length
Replacing normalized
with normalize
normalize
- Returns a variant of a vector normalized by one of Z-score, Min Max, or Mean Max feature scalings.
pow
- raise components of a vector to a given power
join
- concatenate components of a vector into a string
flatten
- Converts a sequence of any of list, tuple, set, int, float into a Vector
sum
- Performs vector addition on a sequence of vectors
New methods: cosinesim
, cost
, mbe
, emul
, ediv
Fixes: Dimensionality Checks