Recipe ratio statistical analysis and comparison tool

A command for calculating mean recipe ratios from recipe data provided in heterageneous units of measure; both volume and weight based.

Also prints the ratio in the form of a recipe ingredient list.

 $ stats sample_input/crepes/swedish_recipe_pannkisar.csv -w 1000 -m milk+water

Recipe ratio in units of weight is 1.00:3.56:1.02:0.17:0.02 (all purpose flour:milk:egg:butter:salt)

1000g Recipe
------------
173g or 329ml all purpose flour
618g or 618ml milk
177g, 150ml or 3 egg(s) where each egg is 53g
29g or 29ml butter
3g or 2ml salt

Note: these calculations are based on 200 distinct recipe proportions. Duplicates have been removed.

$ stats [options] recipe.csv [recipe2.csv]

The CSV files must be of the form,

Flour, Egg, Milk, Butter, Salt
1c, 1 large, 3 cups, 2 tbsp, 0.5 tsp
200g, 55gram, 0.7l, 0, 1 pinch
16oz, 2.5 medium, 2.5c, 1 stick, 0

As you can see, a good deal of freedom is given, so for example "1c" is as good as "1 cup". Weight and volumne based measures can be mixed freely. If an ingredient is missing, simply specifiy 0 without any unit of measure.

If more than one CSV file is given, the column headings must be identical.

-h, --help Prints a summary of the options described below..

-p DIGITS, --precision=DIGITS

Number of digits to show after decimal point for ratio values (default is 2).

-r DIGITS, --recipe-precision=DIGITS

Number of digits to show after decimal point for recipe values (default is 0).

-w GRAMS, --weight=GRAMS

Restrict the total weight to use for the printed recipe. Weight is given in grams (default is 100g).

-v, --verbose

Show confidence intervals and the sample size that is required to reach a confidence interval that is a certain percentage difference from the mean. The default value for the desired difference-from-mean is 5%, but may be adjusted using the --confidence-interval option. Here is an example of how the verbose output looks,

 $ stats sample_input/crepes/swedish_recipe_pannkisar.csv -w 1000 -m milk+water -v

Recipe ratio in units of weight is 1.00:3.56:1.02:0.17:0.02 (all purpose flour:milk:egg:butter:salt)

Recipe ratio with confidence intervals (confidence level is 95%)
----------------------------------------------------------------
The all purpose flour proportion is between 16.68% and 18.00% (the interval is 4% of the mean proportion: 17.34%)
The milk proportion is between 60.67% and 62.84% (the interval is 2% of the mean proportion: 61.75%)
The egg proportion is between 16.98% and 18.49% (the interval is 4% of the mean proportion: 17.74%)
The butter proportion is between 2.40% and 3.38% (the interval is 17% of the mean proportion: 2.89%)
The salt proportion is between 0.24% and 0.32% (the interval is 13% of the mean proportion: 0.28%)

Minimum sample sizes needed for confidence interval with 5% difference and confidence level of 95%
--------------------------------------------------------------------------------------------------
Minimum sample size for all purpose flour proportion: 115
Minimum sample size for milk proportion: 25
Minimum sample size for egg proportion: 146
Minimum sample size for butter proportion: 2300
Minimum sample size for salt proportion: 1399

1000g Recipe
------------
173g or 329ml all purpose flour
618g or 618ml milk
177g, 150ml or 3 egg(s) where each egg is 53g
29g or 29ml butter
3g or 2ml salt

Note: these calculations are based on 200 distinct recipe proportions. Duplicates have been removed.

-i, --include

Include duplicate ingredient proportions from the input data when calculating the recipe statistics. By default duplicates are removed.

-c CONFIDENCE, --confidence-interval=CONFIDENCE

Desired confidence interval expressed as a percentage difference from zero to the mean, default is 0.05 (5%). This option only has effect when the --verbose setting is used to show required sample sizes.

-m MAPPING, --merge=MAPPING

Merge columns either partly or wholly into another. As an example, the following merges the "water" column into the "milk" column,

-m milk+water

The left-most column will be retained. The "water" column, in this case will no longer be visible.

Partial columns may be merged with a specification of the percentage weight to be taken. Take the following example,

-m butter.84+milk.02+cream.2:milk.98+cream.8+butter.16

This merge specification seperates butter fat from three ingredients under the heading "butter" and the remaining liquid under the heading "milk". Notice that a semicolon is used to seperate merge specifications that should result in different columns.

Ingredients with a space in the name can be specified using either quotation marks, or using the column index (starting at zero). The following, for example, are valid specifications,

-m 0+1+2

-m "all purpose flour+granulated sugar"

-t RESTRICTIONS, --restrict=RESTRICTIONS

Restrict individual columns to a given weight. This is useful if, for example, you only have a limited amount of one or more ingredient available. The following example shows how per-ingredient weight restrictions are used,

-t egg=123,sugar=545

Weight is given in grams. With this specification, the printed recipe will not include more than 123g egg or 545g sugar. If this option is used in combination with the --weight option, then the total weight will not exceed the weight specified there. If the --weight option is not used, then the recipe weight will be as high as possible within the given weight restrictions.

As with the column merge option (see above) it is possible to specify ingredient names containing spaces using either the column index (starting at zero) or quotation marks.

-z IGNOREZEROS, --ignore-zeros=IGNOREZEROS

Remove zero values from the calculation of means for one or more columns. The specification is a comma seperated list of column names or indexes. For example, the following specifies that missing values for salt and sugar should be disregarded.

-z salt,sugar

This is useful for ingredients that are frequently missed out from recipes, salt being the prime example.

 $ diff sample_input/crepes/french_recipe_crepes.csv sample_input/crepes/english_recipe_crepes.csv 

Ratio for data set 1 in units of weight is 1.00:1.86:0.12:0.75:0.18:0.01 (all purpose flour:milk:water:egg:butter:salt)
Ratio for data set 2 in units of weight is 1.00:2.17:0.22:1.17:0.23:0.01 (all purpose flour:milk:water:egg:butter:salt)

Percentage difference between salt proportions 58%
Percentage difference between water proportions 40%
Percentage difference between egg proportions 24%
Percentage difference between all purpose flour proportions 20%
Percentage difference between milk proportions 5%
Percentage difference between butter proportions 5%

Overall percentage difference = 25%

$ diff [options] recipe1.csv recipe2.csv [recipe3.csv]

-h, --help Prints a summary of the options described below..

-p DIGITS, --precision=DIGITS

Number of digits to show after decimal point for percentage values (default is 0)

-i, --include

Include duplicate ingredient proportions from the input data when calculating the recipe statistics. By default duplicates are removed.

-c, --change

Show percentage change (percentage difference is default). Here is how the output looks with this option enabled,

 $ diff -c sample_input/crepes/french_recipe_crepes.csv sample_input/crepes/english_recipe_crepes.csv 

Ratio for data set 1 in units of weight is 1.00:1.86:0.12:0.75:0.18:0.01 (all purpose flour:milk:water:egg:butter:salt)
Ratio for data set 2 in units of weight is 1.00:2.17:0.22:1.17:0.23:0.01 (all purpose flour:milk:water:egg:butter:salt)

The salt proportion has increased by 82% from data set 1 to 2
The water proportion has increased by 50% from data set 1 to 2
The egg proportion has increased by 27% from data set 1 to 2
The all purpose flour proportion has decreased by 19% from data set 1 to 2
The butter proportion has increased by 5% from data set 1 to 2
The milk proportion has decreased by 5% from data set 1 to 2

Overall percentage difference = 25%

-m MAPPING, --merge=MAPPING

Same as for stats command (see above).