I pulled two samples of 15 rows each, spread across the dataset: an evenly spaced sample by `Sr No` (100, 400, 600, …, 3000) and a second spot-check sample. Both are in the tables above. Each row is one funding deal, with the startup, date, industry, city, investors, round type, and amount. **What the sample shows** - **Round types:** `Private Equity` and `Seed Funding` dominate the sampled rows. Examples are Medlife (Private Equity) and Dropkaffe (Seed Funding). Some rows also carry a stage hint in `Remarks`, such as "Series A", "Series C", "pre-Series A" or "Bridge funding". - **Cities:** Bangalore, Mumbai, Gurgaon, Chennai and New Delhi are the most common. - **Dates:** the sampled rows run from January 2015 to April 2019. **Data quirks visible in the sample** - **Amounts are text in Indian digit grouping.** `Amount in USD` is a VARCHAR, and "1,70,00,000" means 17,000,000. It must be stripped of commas and cast before any numeric analysis. Many rows are "N/A", for example Digit, The Gourmet Jar and EvenGini. - **Inconsistent spellings.** The same city appears as both "Bengaluru" and "Bangalore". The industry labels also vary in casing ("ECommerce", "eCommerce", "E-commerce"). - **Shifted columns in older rows.** In the 2015 rows (Sr No around 2200 and above, e.g. Razorpay, Sulekha.com), `Industry Vertical` holds a descriptive phrase like "Online Payment Gateway" and `SubVertical` is "nan". These rows look like a different schema from the 2016–2019 ones. - **Missing values are stored as the literal string "nan".** This applies to `Remarks`, `SubVertical` and sometimes `City Location`; Faircent has "nan" for its city, industry and subvertical. - **Messy investor lists.** `Investors Name` is a free-text, comma-separated list. It can have trailing commas ("Accel Partners,") or placeholders like "undisclosed investor". Before any trend or ticket-size analysis, I would normalise the amount column, standardise city and industry labels, and treat "nan" and "N/A" as nulls.