Jay Patel

GPT4o for voice typing ------------------------------------- So far, accuracy with Sellinam, Google voice typing for #tamil is not 100%. After spelled, it requires edit. Tried #GPT4o to output the spoken words in Tamil. It is accurate and doesn't require any hard edit. கேட்பதும் நீ மறுப்பதும் நான் கெஞ்சுவதும் நீ இளகுவதும்... Disclaimer - Haven't tried complex sentences yet. Haven't used any other LLM for voice typing prior.

9

📚 Sentimental Analysis on Twitter Data with Transformer Based Model and Traditional Machine Learning Techniques 1. hashtag #BERT(Bidirectional Encoder Representations from Transformers) 2. hashtag #RoBERTa (Robustly Optimized BERT Approach) 3. hashtag #DistilBERT 4. hashtag #ALBERT 5. hashtag #XLNet 1. hashtag #SVM 2. hashtag #NaiveBayes 3. hashtag #LogisticRegression 4. hashtag #DecisionTree 5. hashtag #RandomForest 6. hashtag #XGBoost 7. hashtag #LightGBM Understanding Sentimental Analysis Sentimental analysis is the process of identifying and categorizing the sentiment expressed in a text. It can range from positive to negative, or neutral. In the context of Twitter data, it's particularly useful for understanding public opinion, tracking brand reputation, and gaining insights into social trends. 💥 Leveraging Transformer-Based Models for Short Text Analysis Transformer-based models, such as BERT, RoBERTa, DistilBERT, and ALBERT, have revolutionized the field of natural language processing (NLP). They excel at capturing the contextual relationships between words in a sentence, making them ideal for short text analysis where subtle nuances can significantly impact sentiment. 1. BERT (Bidirectional Encoder Representations from Transformers): BERT is a bidirectional model that considers the context of a word based on its left and right neighbors. It's a popular choice for many NLP tasks, including sentiment analysis. 2. RoBERTa (Robustly Optimized BERT Approach): RoBERTa is a more robust version of BERT, trained on larger datasets and with different hyperparameters. It often achieves state-of-the-art performance on various NLP benchmarks. 3. DistilBERT: DistilBERT is a smaller, faster version of BERT, making it suitable for resource-constrained environments. It maintains a significant portion of BERT's performance while being more efficient. 4. ALBERT (A Lite BERT): ALBERT is another lightweight alternative to BERT, focusing on reducing the number of parameters without compromising performance. It's designed for efficient training and inference. 💥 Preprocessing Twitter Data Before applying the models, it's essential to preprocess the Twitter data. This typically involves steps like: 1. Tokenization: Breaking down the text into individual words or tokens. 2. Normalization: Converting text to a standard form (e.g., lowercase, removing stop words). 3. Handling special characters: Dealing with emojis, hashtags, and other non-textual elements. #NLP #NaturalLanguageProcessing #MachineLearning #AI #ArtificialIntelligence #SentimentAnalysis #TextClassification #NamedEntityRecognition #MachineTranslation #TextSummarization #QuestionAnswering #DialogSystems #Chatbots #LanguageGeneration hashtag #NLTK #SpaCy #Gensim #Transformers #BERT #GPT #TensorFlow #PyTorch

10

1 Comment

How to get into a data scientist job role for freshers? To become a data scientist as a fresher, you need a strategic approach since most companies don't prefer freshers for this role unless you are from a premium institute like IIT or NIT, or you have strong achievements like a conference paper or winning competitions like Kaggle. Having a strong portfolio is essential. Try to get into other related job roles like software engineer, data engineer, or Python developer, and then switch internally or to other companies. While you are in the team, keep upskilling yourself and work on data science projects that will improve the team's productivity. These impact numbers would help to convince the interview board, and you can easily switch to a data science team. #job #career #datascientist #fresher #ArtificialIntelligence

15

Done #week3 task given by Main Flow Services and Technologies https://lnkd.in/gH7rvciu #week3 #mainflow #internship #python #datavisualization

6

When I did some resume screening in the past I saw few of the patterns like project description saying “100% accuracy achieved” “99% precision achieved” “ Used models like XGBoost , LightGBM for titanic passenger survival prediction” etc I understand that being a fresher fascination with reporting maximum accuracy and working with fancy algos is for real. But those statements are actually spoiling your impression ! - Lets talk about 100% accuracy. Such an accuracy tells that model might have overfitted , there is a possibility of data leakage or data pattern being deterministic in nature. In real industry you hardly encounter such scenarios ! (There are exceptions but those are exceptions only) So if you are flaunting about 100% accuracy may be you are just exposing yourself. (We have worked with such lower values of adj R-squared which we are ashamed of telling 😅) - Going crazy behind fancy models. “ Bro I have always been working with advanced ensembles who does even look at the logistic or linear regression? So old school ! “ No harm in using those models ! But have you checked if similar results could be produced with the help of simple model as well ? Have you identified at what plane of Interpretability vs accuracy does your problem lie ? Do you really think that you will be able to explain this model to a business stakeholder who doesnt even understand abcd of ML ? Is data size enough to use such complex model ? Are you willing to go for such complex system in production ? Believe me as much as XGBoost weighs in ML , in same way Linear/ logistic regressions weighs too (sometimes even more ) Dont think that using Fancy models makes you an advanced data scientist. Your work is to solve business problems Algorithms shouldn’t be picked from a personal bias Choice of algorithm should come from problem and scenario you are dealing with , resources which are there with you etc. Happy to discuss & learn. #machinelearning #datascience #python #learning #interviewpreparation #interview #growth

89

8 Comments

DATA TYPE: --> Numeric Data type --> Floating Data type --> String Data type --> Date and Time Data type Numeric Data type: 1. TINYINT Range: -128 to 127 or 0 to 255 (unsigned) Usage: Best for small numerical values like boolean flags (0, 1). Storage: 1 byte 2. SMALLINT Range: -32,768 to 32,767 or 0 to 65,535 (unsigned) Usage: Suitable for medium-small values such as small identifiers. Storage: 2 bytes 3. MEDIUMINT (Not standard in all SQL databases) Range: -8,388,608 to 8,388,607 or 0 to 16,777,215 (unsigned) Usage: Used for intermediate values where INT is too large; common in MySQL. Storage: 3 bytes 4. INT Range: -2,147,483,648 to 2,147,483,647 or 0 to 4,294,967,295 (unsigned) Usage: Common choice for general numerical data, including user IDs. Storage: 4 bytes 5. BIGINT Range: -9,223,372,036,854,775,808 to 9,223,372,036,854,775,807 or 0 to 18,446,744,073,709,551,615 (unsigned) Usage: Needed for very large numbers, such as global transaction counts. Storage: 8 bytes 6. BIT Range: Typically 0 or 1 Usage: Ideal for storing binary data like boolean values in compact form. Storage: Varies; often bits are packed together. Floating Data type: 1. FLOAT Precision: Up to 7 decimal digits. Usage: Good for storing approximate values where extreme precision isn't critical, like measurements. Storage: Typically 4 bytes. 2. REAL Precision: Similar to FLOAT. Usage: Used interchangeably with FLOAT in many SQL systems; adequate for scientific data that does not require high precision. Storage: Usually 4 bytes. 3. DOUBLE PRECISION Precision: Up to 15 decimal digits. Usage: Essential for high-accuracy requirements such as financial records or precise scientific calculations. Storage: Generally 8 bytes. 4. DECIMAL and NUMERIC Precision: User-defined, with p for the total number of digits and s for digits after the decimal. Example: DECIMAL(10, 2) implies up to 10 total digits, two of which are after the decimal point. Usage: Ideal for accurate and precise calculations, especially in financial contexts where rounding errors are unacceptable. Storage: Depends on the defined precision; generally requires more space than floating-point types to ensure exactness. Next two topic covered in the next post #sql #mysql #database #datatype #numeric #float Oracle MySQL

2