Posted by : at

Category :


principal component analysis(PCA) āĻŦāĻšā§āϞ āĻŦā§āϝāĻŦāĻšāĻŋāϤ Dimensionality Reduction Algorithm. PCA āĻŽā§āϞāϤ āĻāĻ•āϟāĻŋ āĻĄāĻžāϟāĻžāϏ⧇āĻŸā§‡ āĻĄāĻžāϟāĻžāϗ⧁āϞ⧋āϰ Orthogonal Projection ( āϞāĻŽā§āĻŦ āĻ…āĻ­āĻŋāĻ•ā§āώ⧇āĻĒ) āϖ⧁āĻœā§‡ āĻŦ⧇āϰ āĻ•āϰ⧇āĨ¤ Orthogonal projection āĻāϰ āĻŽāĻžāĻ§ā§āϝāĻŽā§‡ PCA āĻĄāĻžāϟāĻžāϏ⧇āĻŸā§‡āϰ āϏāϰāĻŦāĻšā§āĻ› Variance āϖ⧁āĻœā§‡ āĻŦ⧇āϰ āĻ•āϰ⧇, āϜāĻžāϰ āϏāĻžāĻšāĻžāĻ¯ā§āϝ āĻĄāĻžāϟāĻžāϏ⧇āϟ āĻāĻŦāĻ‚ Feature āĻŽāĻ§ā§āϝ⧇ linear-corelation āĻŦ⧇āϰ āĻ•āϰāĻž āϜāĻžāχ!āĨ¤

āφāϰāĻĨāĻžāϤ, āφāĻŽāĻžāĻĻ⧇āϰ āĻ•āĻžāϛ⧇ āϜāĻĻāĻŋ āĻāĻ•āϟāĻŋ āύāĻŋāĻĻāĻŋāĻļāϤ āĻĻāĻžāϤāĻžāϏāĻāϤ⧇āϰ linearlg corelated āĻ•āĻŋāϛ⧁ āĻĢāĻŋāϚāĻžāϰ āĻĨāĻžāϕ⧇ āϤāĻžāχāϞ⧇ PCA āĻāĻ•āϟāĻž suitability orthogonal direction āϖ⧁āĻœā§‡ āĻŦ⧇āϰ āĻ•āϰāϤ⧇ āĻĒāĻžāϰāĻŦ⧇ āϜāĻž āφāĻŽāĻĻ⧇āĻ āĻĻāĻžāϤāĻžāϏ⧇āϰ āĻāϰ āϏāĻŽāϏāĻ¸ā§āϤ āĻĻāĻžāϤāĻž āϕ⧇ āĻāĻ•ā§āϤāĻž direction āĻ āϤ⧁āϞ⧇ āϧāϰāϤ⧇ āĻĒāĻžāϰāĻŦ⧇āĨ¤

PCA āĻ•āϤāϗ⧁āϞāĻŋ principal component āύāĻŋā§Ÿā§‡ āĻ—āĻ āĻŋāϤ, āϚāϞ⧁āύ Principal component āĻ•āĻŋ āĻĻ⧇āϖ⧇ āĻ¨ā§‡ā§Ÿ

Principal component : Principal component āĻšāϞ⧋ Initial Variable ( Raw dataset) āĻĨ⧇āϕ⧇ āĻ¸ā§āϰāĻ¸ā§āϤ Linear combination or mixure āĻāϰ āĻŽāĻžāĻ§ā§āϝāĻŽā§‡ āĻāĻ•āϟāĻŋ New Variable( New Dataset).

New Variable āϟāĻŋ āĻ•āϤāϗ⧁āϞ Principal component āύāĻŋā§Ÿā§‡ āĻ—āϤāĻŋāϤāĨ¤ Principal component āĻāĻ• āĻŦāĻž āĻāĻ•āĻžāϧāĻŋāĻ• āĻšāϤ⧇ āĻĒāĻžāϰ⧇āĨ¤ āĻ…āϰāĻĨāĻžāϤ āĻāĻ•āϟāĻŋ āĻĄāĻžāϤāĻžāϏ⧇āϤ⧇āϰ āĻĄāĻŋāĻŽā§‡āύāĻļāύ āϜāĻĻāĻŋ ā§§ā§Ļā§Ļ āĻšāĻ…āχ āϤāĻŋāĻŦ⧇ āϤāĻžāϰ principal component hobe 100āϤāĻŋāĨ¤ Principal component āϗ⧁āϞ āĻĻāĻžāϤāĻžāϰ āχāύāĻĢāϰāĻŽā§‡āĻļāύ āĻāϰ āωāĻĒāϰ⧇ āĻ­āĻŋāĻ¤ā§āϤāĻŋ āĻ•āϰ⧇ āύāĻŋāĻŽā§āύāĻ•ā§āϤ āĻŦāĻŋāύāĻžāϏ āφāĻ•āĻžāϰ⧇ āϏāĻžāϜāĻžāύ⧇āĻž āĻĨāĻžāϕ⧇ â€Ļâ€Ļâ€Ļâ€Ļ.

New variable āϟāĻŋ totally uncorrelated āĻšā§Ÿā§‡ āĻĨāĻžāϕ⧇ āĻāĻŦāĻ‚ Initail variable āĻāϰ āĻ…āϧāĻŋāĻ•āĻžāĻ‚āĻļ āχāύāĻĢāϰāĻŽā§‡āĻļāύ compressed āĻšā§Ÿā§‡ 1st pricipal component create kore thake.

PCA try āĻ•āϰ⧇ āĻ…āϧāĻŋāĻ•āĻžāĻ‚āĻļ āύāĻŋāϰāĻ­āϰ āχāύāĻĢāϰāĻŽā§‡āĻļāύ 1st principal component āĻ āϰāĻžāĻ–āĻžāϰ āϤāĻžāϰāĻĒāϰ āĻ…āĻŦāĻļāĻŋāĻˇā§āϟ āĻ…āϧāĻŋāĻ•āĻžāĻ‚āĻļ āχāύāĻĢāϰāĻŽā§‡āĻļāύ 2nd principal component āĻ āϰāĻžāĻ–āĻžāϰ āĻāĻŦāĻ‚ āĻāχāĻ­āĻžāĻŦ⧇ āχāύāĻĢāϰāĻŽā§‡āĻļāύ āĻāϰ āωāĻĒāϰ⧇ āĻ­āĻŋāĻ¤ā§āϤāĻŋ āĻ•āϰ⧇ principal component āĻāϰ āĻŦāĻŋāύāĻžāϏ create āĻšā§Ÿā§‡ āĻĨāĻžāϕ⧇āĨ¤ āύāĻŋāĻŽā§āύ⧇ āĻ›āĻŋāĻ¤ā§āϰ⧇ āĻĻ⧇āĻ–āĻžāύ⧋ āĻšāϞ⧋āĨ¤ picture hereâ€Ļâ€Ļ

āωāĻĒāϰāĻŋāωāĻ•ā§āϤ āĻŦāĻŋāĻ¨ā§āύāĻžāϏ(higher information to lower information) āφāĻ•āĻžāϰ⧇ principal components āϗ⧁āϞāĻž āϏāĻžāϜāĻŋā§Ÿā§‡ āϖ⧁āĻŦ āϏāĻšāĻœā§‡ āφāĻŽāϰāĻž āĻ•āĻŽ āχāύāĻĢāϰāĻŽā§‡āĻļāύ āύāĻˇā§āϟ āĻ•āϰ⧇ āĻāĻ•āϟāĻŋ Lower dimensional Dataset( new Dataset) create hoi.

āϏ⧁āϤāϰāĻžāĻ‚, āĻāχāĻ­āĻžāĻŦ⧇ lower information principal component āĻŦāĻžāĻĻ āĻĻāĻŋā§Ÿā§‡ āĻ…āĻŦāĻļāĻŋāĻ¸ā§āϟ principal component āύāĻŋā§Ÿā§‡ initail varaiable(Raw Dataset) āĻĨ⧇āϕ⧇ new Variable ( new Dataset) create hoye thake.

Example : āφāĻŽāϰāĻž āϜāĻžāύāĻŋ āĻāĻ•āϟāĻž āĻĻāĻžāϤāĻžāϏ⧇āϤ āĻāϰ dimension jodi 100D hoye tobe tar principal component o hobe 100ti. PCA jokhon dimension reduction kore tokhon low variance feature ke bad diye higher dimension theke lower dimension dataset create kore thake.. Orthat optimal principal component khuje ber korar jonno PCA sob somoi low information feature or low variance data ke noise hisabe bibecona kore. Ei noise feature gula PCA bad diye ekta notun dataset create kore thake jar dimension hoii Main dataset er dimension theke onkk kom ( deoend on infomation gather by each princiapl component).

Dhoren main datasert er name A. A datser er dimension hosce 100D ebong er principal component o 100ti. 100 ti principal component er mjhee 1st 20 principal component e 95% data information hold kore.

PCA tokhon 21-100 porjonto dimension er data ke noise hisabe bibchone kore ogula reomve kore dibe. Baki 20 principle component niye ekta new Dataset create kore B.

sutrang PCA apply kore, B dataset ti A dataset 95% information hold kore, higher dimensional dataset (A: 100 dimension) theke lower dimensional dataset( B : 20 dimension) create korbe.

Note :: Dekha gese noise feature or low variance data gula suoervied

About

Hello, My Name is Jahid Hasan. I love to Code and play with robotics and AI....

Star