Implementation of KNN and comparison with scikit-learn built-in KNN

Project

Football Player Transfer Prediction

Football Player Transfer Prediction Using Different Classifiers

Project Report : Football Player Transfer Prediction Report

Project : Football Player Transfer Prediction Using Different Classifiers

DataSet : LINK

Football Player Transfer Dataset

Dataset Documention: Football Player Transfer Dataset Documention

Dataset: Football Player Transfer Dataset

Computer Store Management System Without UI In PL/SQL

Project Report by Nafis Islam on Scribd

Project Link

Use of CFGs for Parsing OPEN

We can think of using CFGs to detect various language constructs in the token streams freed from simple
syntactic and semantic errors, as it is easier to describe the constructs with CFGs. But CFGs are hard to
apply practically. In this session we first exercise on simpler implementations of CFGs, and then
implement the FIRST and FOLLOW functions that facilitate efficient parsing algorithms for practical
problems.

Session 5 by Nafis Islam on Scribd

Detecting Simple Syntax Errors

Syntax errors are very common in source programs. Suppose, a given C source program has been scanned, filtered, lexically analyzed and tokenized as that were done in earlier sessions. In addition, line numbers have been assigned to the source code lines for generating proper error messages. As the first step to Syntax Analysis, we now perform detection of simple syntax errors like duplication of tokens except parentheses or braces, Unbalanced braces or parentheses problem, unmatched ‘else’ problem, etc

Session 4 by Nafis Islam on Scribd

Symbol Table Generation

A given C source program has been scanned, filtered and then lexically analyzed as it was done in Session 2. We have all the lexemes marked as different types of tokens like keywords, identifiers, operators, separators, parentheses, numbers, etc. We now get corrected the unknown lexemes first, and then generate a Symbol Table describing the features of the identifiers. Finally, we generate a modified token stream in accordance with the Symbol Table for processing by the next phase, that is, Syntactic Analysis

Session 3 by Nafis Islam on Scribd

Lexical Analysis

we have a C source program scanned and filtered as it was done in Session 1. We now take that modified file as input, and separate the lexemes first. We further recognize and mark the lexemes as different types of tokens like keywords, identifiers, operators, separators, parenthesis, numbers, etc.

Session 2 by Nafis Islam on Scribd

Scanning and Filtering a Source Program

You are given a C source program with single and multiple line comments. As the first step toward compilation you need to remove the comments and white space (extra spaces, tabs and newline characters). Develop a program that takes as input file the given source program and produces a filtered file as stated above. The program must also display both the files.

Session 1 by Nafis Islam on Scribd

A Simple Presentation On Particle Swarm Optimization

The Particle Swarm Optimization by Nafis Islam on Scribd

Into to Meta-Heuristics

Meta-heuristic হল একটি primary field of stochastic optimization.

Stochastic optimization হল the general class of algorithm and techniques যা randomness use করে optimal বা optimal এর কাছাকাছি solution বের করে আনে।

•যে সকল problem এ আমরা আগে থেকে  বুঝতে পারি না যে, optimal solution কেমন হতে পারে।

•যে সকল problem এ আমরা জানি না, principal way তে  কিভাবে solution এর দিকে আগাব।

•যে সকল problem এ, brute-force search impossible কারন search space অনেক বড়।


        কিন্তু  আমাদের কাছে ঐ  problem এর কোন candidate solution দিলে   test করে দেখতে পাড়ব solution তা কত ভাল।সে সকল problem এ meta-heuristic use করা হয়।

“Meta-heuristics are applied to I know it when I see it problem”

        এরকম কোন problem এর সামনে পরলে সবচেয়ে simplest thing যা করতে পারি তা হল  random search, আমরা random solution বের করব যতক্ষণ আমাদের কাছে  time আছে, সবচেয়ে best solution return করব।

কিন্তু random search এর আগে একটা  alternative পথ দেখতে পারি।

      আমরা প্রথমে random solution নেই। এই solution একটা ছোট  এবং random modification করি। এখন নতুন solution test করি।
যদি দেখি current solution better হয় then old solution ফেলে দিব। যদি না হয় তাহলে current solution ফেলে দিব।
     এখন বর্তমান হাতে যে solution আসে  তার উপর ছোট ও random modification করব। এভাবে যতক্ষণ পারা যায় repeat করতে থাকব।
     এই পধতি কে বলে hill-climbing.এটি হল একটি simple meta-heuristic algorithm.

    "It exploits a heuristic belief about your space of candidate solutions which is usually true for many problems: that similar solutions tend to behave similarly (and tend to have similar quality), so small modifications will generally result in small, well-behaved changes in quality, allowing us to “climb the hill” of quality up to good solutions. This heuristic belief is one of the central defining features of meta-heuristics: indeed, nearly all meta-heuristics are essentially elaborate combinations of hill-climbing and random search."

     Meta-heuristic দারা যে সকল problem tackle করা যায় সে সকল problem হল sub-class of inverse problem। 

       Inverse problem কে আমারা এভাবে কল্পনা করতে পাড়ি, মনে করি আমাদের কাছে একটি test function আছে  f() যেটা parameter হিসেবে একটি candidate solution কে নিবে এবং ওই solutionটির একটি assessment বা fitness return করবে।

       কিন্তু এটা মোটামোট impossible উপরের functionটির inverse function f-1() তৈরি করা যা assessment বা fitness parameter হিসেবে নেয় এবং এমন একটি candidate solution return করবে যার ঐ assessment or fitness থাকবে।

      Optimization methods(Meta-heuristic) design করা হয়েছে  inverse problem কে overcome করতে।