AI:Corporate Identity, Poltical, TTechnoocratz

 

Here are the Prime Ministers of India ranked by total time in office (approximate durations):

Prime Minister	Time in Office
Jawaharlal Nehru	16 years, 286 days
Indira Gandhi	15 years, 350 days (across two terms)
Narendra Modi	4,399+ days (about 12 years as of 10 June 2026, and continuing)
Manmohan Singh	10 years, 4 days
Atal Bihari Vajpayee	6 years, 80 days
Rajiv Gandhi	5 years, 32 days
P. V. Narasimha Rao	4 years, 330 days
Morarji Desai	2 years, 126 days
Lal Bahadur Shastri	1 year, 216 days
V. P. Singh	343 days
Chandra Shekhar	223 days
H. D. Deve Gowda	324 days
I. K. Gujral	332 days
Charan Singh	170 days
Vishwanath Pratap Singh	343 days
Gulzarilal Nanda	26 days (acting PM in two brief periods)

Current record holders:

• Longest-serving PM overall: Jawaharlal Nehru — 16 years, 286 days. 

• Longest-serving continuously elected PM: Narendra Modi — 4,399+ consecutive days as of 10 June 2026. 

If you'd like, I can also give the exact number of days served by every Prime Minister in a simple list.

If Ambani Adani are there and Arvind Krishna ibm us is there after narendra Modi then who’re similar entities in corporate world technocratz in all above prime ministers tenure 

If you're asking, "During each Prime Minister's era, who were the most influential business leaders or technocrats in India (similar to how today we think of Ambani, Adani, or IBM CEO Arvind Krishna)?", that's somewhat subjective, but a reasonable historical mapping would be:

Prime Minister Era	Prominent Business / Technology Figures
Jawaharlal Nehru (1947–1964)	J. R. D. Tata, G. D. Birla, Homi J. Bhabha
Lal Bahadur Shastri (1964–1966)	J. R. D. Tata, Vikram Sarabhai
Indira Gandhi (1966–1977, 1980–1984)	J. R. D. Tata, Dhirubhai Ambani, Satish Dhawan
Morarji Desai / Charan Singh	Dhirubhai Ambani, J. R. D. Tata
Rajiv Gandhi (1984–1989)	Sam Pitroda, Azim Premji, Dhirubhai Ambani
V. P. Singh, Chandra Shekhar	Azim Premji, Dhirubhai Ambani
P. V. Narasimha Rao (1991–1996)	N. R. Narayana Murthy, Azim Premji, Ratan Tata
H. D. Deve Gowda / I. K. Gujral	N. R. Narayana Murthy, Azim Premji
Atal Bihari Vajpayee (1998–2004)	N. R. Narayana Murthy, Azim Premji, Ratan Tata
Manmohan Singh (2004–2014)	Mukesh Ambani, Ratan Tata, Nandan Nilekani
Narendra Modi (2014–present)	Mukesh Ambani, Gautam Adani, Arvind Krishna, Sundar Pichai, Satya Nadella

A broader pattern is:

• 1947–1970s: Tata, Birla, and national scientists such as Bhabha and Sarabhai dominated influence.

• 1980s: Dhirubhai Ambani emerged as the most disruptive private-sector entrepreneur.

• 1990s: Liberalization elevated Infosys, Wipro, and technology leaders such as Narayana Murthy and Azim Premji.

• 2000s: Ratan Tata, Mukesh Ambani, and Nandan Nilekani became nationally influential.

• 2010s–2020s: Mukesh Ambani, Gautam Adani, and globally prominent Indian-origin tech CEOs such as Arvind Krishna, Sundar Pichai, and Satya Nadella became the most visible corporate and technocratic figures.

There isn't a single exact equivalent for every PM era, but J. R. D. Tata (1950s–1980s), Dhirubhai Ambani (1980s), Narayana Murthy (1990s), Ratan Tata (2000s), and Mukesh Ambani (2010s–2020s) are often viewed as the defining corporate figures of their respective periods.

And similar list with prime ministers across US Canada UK Africa Middle East Russia China Korea Asia and Iceania 

If you want a global version—pairing major political leaders with the most influential business leaders/technocrats of roughly the same era—here's a high-level historical view. These are not official pairings; they are people who had outsized economic, technological, or industrial influence during those periods.

United States

Political Era	Influential Corporate / Technology Figures
Dwight D. Eisenhower	Thomas J. Watson Jr.
John F. Kennedy	David Rockefeller
Ronald Reagan	Steve Jobs, Bill Gates
Bill Clinton	Bill Gates, Andy Grove
Barack Obama	Elon Musk, Jeff Bezos, Tim Cook
Donald Trump / Joe Biden	Elon Musk, Jensen Huang, Satya Nadella

United Kingdom

Political Era	Influential Corporate / Technology Figures
Margaret Thatcher	Richard Branson
Tony Blair	Richard Branson, Martin Sorrell
Recent PMs	Demis Hassabis, James Dyson

Canada

Political Era	Influential Corporate / Technology Figures
Pierre Trudeau	Kenneth Thomson
Jean Chrétien	Jim Balsillie, Mike Lazaridis
Justin Trudeau	Tobi Lütke

Russia / Soviet Union

Political Era	Influential Corporate / Technology Figures
Nikita Khrushchev	Sergei Korolev
Mikhail Gorbachev	Sergei Pugachev
Vladimir Putin	Alexey Miller, German Gref

China

Political Era	Influential Corporate / Technology Figures
Deng Xiaoping	Ren Zhengfei
Jiang Zemin	Ren Zhengfei
Hu Jintao	Jack Ma
Xi Jinping	Pony Ma, Lei Jun

Japan

Political Era	Influential Corporate / Technology Figures
Post-war economic boom	Akio Morita, Soichiro Honda
1980s–1990s	Akio Morita
Recent decades	Tadashi Yanai, Masayoshi Son

South Korea

Political Era	Influential Corporate / Technology Figures
Park Chung-hee	Lee Byung-chul
Democratization era	Chung Ju-yung
Modern era	Lee Jae-yong

Middle East

Saudi Arabia

• Mohammed bin Salman → Amin H. Nasser

United Arab Emirates

• Mohammed bin Rashid Al Maktoum → Sultan Ahmed bin Sulayem

Israel

• David Ben-Gurion → Stef Wertheimer

• Benjamin Netanyahu → Eyal Waldman

Africa

South Africa

• Nelson Mandela → Nicky Oppenheimer

• Recent era → Elon Musk (global influence)

Nigeria

• Olusegun Obasanjo → Aliko Dangote

Egypt

• Hosni Mubarak → Naguib Sawiris

Oceania

Australia

• Bob Hawke → Rupert Murdoch

• Recent era → Andrew Forrest, Mike Cannon-Brookes

New Zealand

• Jacinda Ardern → Peter Beck

The broad "corporate dynasty" timeline

If you compress world business history into a few dominant figures by era, a rough sequence is:

J. R. D. Tata / G. D. Birla (1950s–60s) → Dhirubhai Ambani (1980s) → Bill Gates (1990s) → Steve Jobs & Jeff Bezos (2000s) → Elon Musk, Mukesh Ambani, Jack Ma (2010s) → Jensen Huang, Satya Nadella, Sundar Pichai, Arvind Krishna, Gautam Adani (2020s).

These individuals played roles in business and technology influence that, in their respective eras, can be compared to the political prominence of major prime ministers and presidents.

7 Scikit-learn Tricks for Optimized Cross-Validation

 

7 Scikit-learn Tricks for Optimized Cross-Validation
Image by Editor | ChatGPT

Introduction

Validating machine learning models requires careful testing on unseen data to ensure robust, unbiased estimates of their performance. One of the most well-established validation approaches is cross-validation, which splits the dataset into several subsets, called folds, and iteratively trains on some of them while testing on the rest. While scikit-learn offers standard components and functions to perform cross-validation the traditional way, several additional tricks can make the process more efficient, insightful, or flexible.

This article reveals seven of these tricks, along with code examples of their implementation. The code examples below use the scikit-learn library, so make sure it is imported.

I recommend that you first acquaint yourself with the basics of cross-validation by checking out this article. Also, for a quick refresher, a basic cross-validation implementation (no tricks yet!) in scikit-learn would look like this:

from sklearn.datasets import load_iris from sklearn.model_selection import cross_val_score from sklearn.linear_model import LogisticRegression   X, y = load_iris(return_X_y=True)   model = LogisticRegression(max_iter=200)   # Basic cross-validation strategy with k=5 folds scores = cross_val_score(model, X, y, cv=5)   # Cross validation results: per iteration + aggregated print("Cross-validation scores:", scores) print("Mean score:", scores.mean())

The following examples assume that the basic libraries and functions, like cross_val_score, have already been imported.

1. Stratified cross-validation for imbalanced classification

In classification tasks involving imbalanced datasets, standard cross-validation may not guarantee that the class proportions are represented in each fold. Stratified k-fold cross-validation addresses this challenge by preserving class proportions in each fold. It is implemented as follows:

from sklearn.model_selection import cross_val_score, StratifiedKFold   cv = StratifiedKFold(n_splits=5) scores = cross_val_score(model, X, y, cv=cv)

2. Shuffled K-fold for Robust Splits

By using a KFold object along with the shuffle=True option, we can shuffle instances in the dataset to create more robust splits, thereby preventing accidental bias, especially if the dataset is ordered according to some criterion or the instances are grouped by class label, time, season, etc. It is very simple to apply this strategy:

from sklearn.model_selection import KFold   cv = KFold(n_splits=5, shuffle=True, random_state=42) scores = cross_val_score(model, X, y, cv=cv)

3. Parallelized cross-validation

This trick improves computational efficiency by using an optional argument in the cross_val_score function. Simply assign n_jobs=-1 to run the process at the fold level on all available CPU cores. This can result in a significant speed boost, especially when the dataset is large.

scores = cross_val_score(model, X, y, cv=5, n_jobs=-1)

4. Cross-Validated Predictions

By default, using cross-validation in scikit-learn yields the accuracy scores per fold, which are then aggregated into the overall score. If instead we wanted to get predictions for every instance to later build a confusion matrix, ROC curve, etc., we can use cross_val_predict as a substitute for cross_val_score, as follows:

from sklearn.model_selection import cross_val_predict   y_pred = cross_val_predict(model, X, y, cv=5)

5. Beyond Accuracy: Custom Scoring

It is also possible to replace the default accuracy metric used in cross-validation with other metrics like recall or F1-score. It all depends on the nature of your dataset and your predictive problem’s needs. The make_scorer() function, along with the specific metric (which must also be imported), achieves this:

from sklearn.metrics import make_scorer, f1_score, recall_score   f1 = make_scorer(f1_score, average="macro") # You can use recall_score too scores = cross_val_score(model, X, y, cv=5, scoring=f1)

6. Leave One Out (LOO) Cross-Validation

This strategy is essentially k-fold cross-validation taken to the extreme, providing an exhaustive evaluation for very small datasets. It is a useful strategy mostly for building simpler models on small datasets like the iris one we showed at the beginning of this article, and is generally not advisable for larger datasets or complex models like ensembles, mainly due to the computational cost. For a little extra boost, it can be optionally used combined with trick number #3 shown earlier:

from sklearn.model_selection import LeaveOneOut   cv = LeaveOneOut() scores = cross_val_score(model, X, y, cv=cv)

7. Cross-validation Inside Pipelines

The last strategy consists of applying cross-validation to a machine learning pipeline that encapsulates model training with prior data preprocessing steps, such as scaling. This is done by first using make_pipeline() to build a pipeline that includes preprocessing and model training steps. This pipeline object is then passed to the cross-validation function:

from sklearn.pipeline import make_pipeline from sklearn.preprocessing import StandardScaler   pipeline = make_pipeline(StandardScaler(), LogisticRegression(max_iter=200)) scores = cross_val_score(pipeline, X, y, cv=5)

Integrating preprocessing within the cross-validation pipeline is crucial for preventing data leakage.

Wrapping Up

Applying the seven scikit-learn tricks from this article helps optimize cross-validation for different scenarios and specific needs. Below is a quick recap of what we learned.

Trick	Explanation
Stratified cross-validation	Preserves class proportions for imbalanced datasets in classification scenarios.
Shuffled k-fold	By shuffling data, splits are made more robust against possible bias.
Parallelized cross-validation	Uses all available CPUs for boosting efficiency.
Cross-validated predictions	Returns instance-level predictions instead of scores by fold, useful for calculating other metrics like confusion matrices.
Custom scoring	Allows using custom evaluation metrics like F1-score or recall instead of accuracy.
Leave One Out (LOO)	Thorough evaluation suitable for smaller datasets and simpler models.
Cross-validation on pipelines	Integrates data preprocessing steps into the cross-validation process to prevent data leakage.