Data & method

Where the food data comes from

The 246 core dishes come from the ThaliFacts dataset, which is built on the Indian Food Composition Tables 2017 published by the ICMR National Institute of Nutrition, Hyderabad, combined with standard recipe analysis for composite dishes. Using the same dataset as ThaliFacts is deliberate: a roti has to be the same number of calories on both sites or neither site is worth reading.

The 37 dishes estimated here

A Bengali, Odia or Assamese plan cannot be built from that dataset alone, because several everyday dishes in those kitchens are not in it. Rather than substitute a north Indian dish and pretend, 37 regional dishes were estimated specifically for this site by recipe analysis: the standard household recipe is broken into ingredients, each ingredient is valued from IFCT 2017, and the result is divided by the cooked yield.

These are estimates and are marked est everywhere they appear. Their energy figures are stated as 4P + 4C + 9F precisely so they can never disagree with their own macronutrients.

How a meal total is computed

Each dish has a stated serving in grams. A meal lists dishes and a number of servings. Grams = serving weight × servings; nutrients scale linearly from the per-100 g figures. A meal total is the sum of its dishes, a day total is the sum of its five meals, and the weekly average is the sum of the seven days divided by seven. There is no adjustment factor anywhere and nothing is rounded before the sum — the rounding you see happens only at display.

Every build runs an Atwater reconciliation over all 283 foods: 4 kcal per gram of protein, 4 per gram of available carbohydrate and 9 per gram of fat must land within 15% of the stated energy (20% for very high-fibre foods, where fibre accounting differs). The build refuses to produce a site if any food fails.

How the plans are put together

Every meal in the library is a real combination of real dishes; none of them was designed to hit a number. The build then selects which meals go on which day so the week lands inside the plan's stated calorie band and clears its protein and fibre floors, while avoiding the same meal appearing on consecutive days. It never scales a portion or shaves a figure to make a total work. If a plan's stated average does not equal the sum of its own days divided by seven, that is a bug and the verification step fails the build.

How the grocery quantities are derived

Each dish carries a documented raw-ingredient mapping for one serving — one roti is 30 g of atta, one katori of dal is 30 g of dry dal, one katori of curd is 150 g, one katori of chicken curry is 85 g of raw chicken. The plan knows how many servings of each dish it contains, so multiplying through gives a week's shopping. Fresh vegetables are pooled into one line, because nobody shops for 110 g of bhindi.

Indicative rates used for the rupee figures

These are coarse mid-2026 metro retail rates set for scale, not a price index. They are not sourced from a published series and should not be quoted as one. Prices swing 30% or more between cities, seasons and shop types, and buying dal by the kilo rather than in a branded packet moves them further. The quantities are the useful part of a grocery list; the rupee column is context.

How the plate quadrant chart is portioned

The chart above every plan reads the same week a second way. Instead of asking what a day totals, it asks what a day looks like: how much of the plate is grain, how much is dal, how much is vegetable, and how far each of those is from the quarter marks a clinic handout draws.

Every dish in the library is assigned to exactly one of the 5 groups below. Composite dishes — khichdi, masala dosa, sambar, misal — are assigned whole, to the group that dominates the serving by weight. They are not split, because splitting them would need a recipe breakdown this site does not have, and inventing one to make a diagram tidier is exactly the move the rest of this page refuses. The build fails if any dish used in a plan is not in the table below.

The outer ring is each group's share of the day by served weight; the inner ring is its share by energy. The two disagree on purpose — that disagreement is the only thing on the page a calorie total cannot say. Portions are printed in katori of 150 g, the same katori used by the serving weights and the grocery mapping, so one katori means one thing everywhere on this site.

What the chart does not claim: that these five groups are a dietary guideline, that the quarter marks are a target for you, or that a group's share should be any particular number. They are a ruler, not a verdict.

What this method cannot do

• It cannot know your kitchen. Fat is the largest source of variance in Indian home cooking by a wide margin. Every figure assumes about a teaspoon of oil per person in a sabzi. Double it and a day gains roughly 200 kcal with nothing new on the plate.
• It cannot know your portions. A katori is taken as 150 g and a phulka as 45 g. If your roti is a third bigger, so is every roti figure on this site.
• Restaurant food is a different animal. The same dish out can run 30 to 60% higher.
• Composition tables report averages. Individual batches, varieties and seasons vary.
• It cannot know you. Requirements change with age, activity, pregnancy, medication and diagnosed conditions. That is what a registered dietitian is for.

Corrections

If a figure here disagrees with IFCT 2017, IFCT wins and the page gets fixed. If one of the 37 estimates above is wrong and you can say why, that is the most useful email this site can receive. Send the page URL and what looks wrong.

Every reference behind this page, with its publication date and the month it was last checked, is on the sources page — including the one column on this site that has no published source at all.

General information only — not medical or nutrition advice. These are illustrative plans built on average values; they are not prescriptions and cannot account for allergies, conditions or medication.

ਸਰੋਤ·Computed from ThaliFacts food data·ਸਮੀਖਿਆ ਜੁਲਾਈ 2026