Weight tracking scripts

calc

#!/bin/sh

#set -e

./ewma.py < log > smoothed
echo
./progress.py "$@" < smoothed
echo

gnuplot < weight-graph.gpl
#./maintain.py < log
#echo

./rate.py < smoothed > rate
gnuplot < rate-graph.gpl

cp log backup/$(tail -n 1 log | awk '{print $1}')

ewma.py

#!/usr/bin/python
import sys
import argparse
import time
import datetime

argparser = argparse.ArgumentParser()
argparser.add_argument('--verbose', '-v', action='store_true')
argparser.add_argument('--alpha', '-a', type=float, default=0.1)
args = argparser.parse_args()

if args.alpha < 0.0 or args.alpha > 1.0:
    raise ValueError, "you suck"

ewma = 0.0
def update(weight):
    global ewma
    ewma = args.alpha*weight + (1.0 - args.alpha)*ewma

last_ordinal = None
last_weight = None
for ln in sys.stdin:
    date, weight = ln.split()
    date = datetime.datetime.strptime(date, '%Y-%m-%d')
    weight = float(weight)
    ordinal = date.toordinal()

    if args.verbose:
        print (date.toordinal(), weight)

    if last_ordinal is None:
        last_ordinal = ordinal
        last_weight = weight
        ewma = weight
    else:
        last_ordinal += 1
        while last_ordinal < ordinal:
            if args.verbose:
                print 'Missing day, using', last_weight
            update(last_weight)
            last_ordinal += 1

    if args.verbose:
            print 'Update with', weight
    update(weight)
    last_weight = weight

    print date.strftime('%Y-%m-%d'), ewma

maintain.py

#!/usr/bin/env python
import datetime
import sys
import math
import argparse
import numpy
from scipy import stats

goal = 234
window = 30
points = []
confidence = 0.05

for ln in sys.stdin:
    date, weight = ln.split()
    date = datetime.datetime.strptime(date, '%Y-%m-%d')
    weight = float(weight)
    points.append((date, weight))
    points = points[-window:]

weights = [p[1] for p in points]
minw, maxw, avgw = (min(weights), max(weights), numpy.average(weights))

t_stat, p_value = stats.ttest_1samp(weights, goal)

print 'Target:         %6.2f lbs' % (goal,)
print 'Range:          %6.2f lbs' % (maxw-minw,)
print 'Std deviation:  %6.2f lbs' % (numpy.std(weights),)
print 'Min/avg/max:    %6.2f  /  %6.2f  /  %6.2f' % (minw, avgw, maxw)
print 'Deltas:         %6.2f  /  %6.2f  /  %6.2f' % (minw-goal, avgw-goal, maxw-goal)
print
print 'Verdict:       ', 'Maintaining!' if p_value > confidence else 'Not maintaining', '(p = %.2f)' % (p_value,)

progress.py

#!/usr/bin/env python
import datetime
import sys
import math
import argparse
from scipy import stats

argparser = argparse.ArgumentParser()
argparser.add_argument('--reddit', '-r', action='store_true')
args = argparser.parse_args()

parse_date = lambda s: datetime.datetime.strptime(s, "%Y-%m-%d")

goal = 225
goal_date = parse_date("2019-06-01")
window = 30
height = 1.91  # meters

kg_per_lb = 0.453592
points = []

def parse(ln):
    date, weight = ln.split()
    date = datetime.datetime.strptime(date, '%Y-%m-%d')
    weight = float(weight)
    return (date, weight)

lines = list(sys.stdin)
start_date, start = parse(lines[0])

points = map(parse, lines)
num_points = len(points)
num_decrease = sum(1 for p, q in zip(points, points[1:]) if p[1] > q[1])

points = points[-window:]

first = points[0]
last  = points[-1]

reg_x = [p[0].toordinal() for p in points]
reg_y = [p[1] for p in points]

slope, intercept, r_value, p_value, stderr = stats.linregress(reg_x, reg_y)

#rate = (first[1] - last[1]) / (last[0] - first[0]).days
rate = -slope

if rate <= 0.0:
    print "Sorry, you're getting fatter at %.2f lb/wk (r^2 = %.2f)" % (-7.0*rate, r_value**2.0)
    sys.exit(1)

def output_stat(label, value):
    if args.reddit:
        print '**' + label + '**|' + value.lstrip()
    else:
        print (label + ':').ljust(17) + value

done = start - last[1]
days_done = (last[0] - start_date).days
remainder = last[1] - goal
remaining_time = datetime.timedelta(days = remainder / rate)
finish_date = last[0] + remaining_time
margin = (goal_date - finish_date).days + 1

required_rate = remainder / (goal_date - last[0]).days
excess_rate = rate - required_rate

def project(date):
    projection = last[1] - (rate * (parse_date(date) - last[0]).days)
    print 'Weight on %s:   %6.2f lbs = %6.2f lbs lost' % (date, projection, start - projection)


if args.reddit:
    print 'Stat|Value'
    print ':-|:-'
output_stat('Current weight', '%6.2f lbs' % (last[1],))
#output_stat('BMI', '%6.2f' % ((kg_per_lb * last[1]) / (height ** 2.0),))
output_stat('Lost so far', '%6.2f lbs = %5.2f%% of starting weight' % (done, 100.0 * done / start))
output_stat('Remaining', '%6.2f lbs = %5.2f%% of current  weight' % (remainder, 100.0 * remainder / last[1]))
output_stat('Progress', '%6.2f %%' % (done / (start - goal) * 100.0,))
output_stat('Required rate', '%6.2f lbs / wk' % (7.0 * required_rate,))
output_stat('Actual rate', '%6.2f lbs / wk = %.2f%% per week  (r^2 = %4.2f)' % (7.0*rate, 100*7.0*rate / last[1], r_value**2.0))
#output_stat('Excess rate', '%6.2f lbs / wk' % (7.0*excess_rate,))
output_stat('Deficit', '  %4d kcal / day' % (rate*3500,))
#output_stat('Excess deficit', '  %4d kcal/day' % (excess_rate*3500,))
output_stat('All-time rate', '%6.2f lbs / wk' % (7.0 * done / days_done,))
output_stat('Average drops', '%6.2f %% of the time (%d out of %d points)' % (100.0 * num_decrease / float(num_points), num_decrease, num_points))
print
print 'Goal will be reached on %s' % (finish_date.strftime('%Y-%m-%d'),),
if margin > 0:
    print '(%d days early)' % (margin,)
else:
    print '(%d days late)' % (-margin,)
if args.reddit:
    print
print days_done, 'days done, only', remaining_time.days, 'left to go! Keep it up, you\'re doing great!'

print
project('2019-06-01')

with open('regression_line', 'w') as f:
    line_out = lambda pt: f.write(pt[0].strftime('%Y-%m-%d') + ' ' + str(intercept + slope*pt[0].toordinal()) + '\n')
    line_out(points[0])
    line_out(points[-1])

rate-graph.gpl

set title "Weight loss rate"
set ylabel "Pounds per week"
#set y2label "r^2"

set term png size 1400,900
set output "rate-graph.png"
set timefmt "%Y-%m-%d"
set xdata time
set format x "%b\n%d"

#set xrange ["2018-02-18":]

plot "rate" using 1:2 with lines lw 3 title "Rate", \
    #"rate" using 1:3 with lines lw 1 title "r^2" axes x1y2

rate.py

#!/usr/bin/env python
import datetime
import sys
import math
import argparse
from scipy import stats

window = 14
points = []

for ln in sys.stdin:
    date, weight = ln.split()
    date = datetime.datetime.strptime(date, '%Y-%m-%d')
    weight = float(weight)
    points.append((date, weight))
    points = points[-window:]

    if len(points) < window:
        continue

    reg_x = [p[0].toordinal() for p in points]
    reg_y = [p[1] for p in points]

    slope, intercept, r_value, p_value, stderr = stats.linregress(reg_x, reg_y)

    print date.strftime("%Y-%m-%d"), -7.0*slope, r_value**2.0

weight-graph.gpl

set title "Weight loss progress"
set ylabel "Weight (lbs)"

set term png size 1400,900
set output "weight-graph.png"
set timefmt "%Y-%m-%d"
set xdata time
set format x "%b\n%d"

#set xrange ["2018-08-01":]
#set yrange [240:290]

plot "log" using 1:2 with linespoints pointtype 6 title "Daily weight", \
     "smoothed" using 1:2 with lines lw 3 title "Moving average", \
     "regression_line" using 1:2 title "Trend line" with lines lw 2 lt rgb "#c01f1f", \
#     "raw_line" using 1:2 title "Raw trend line" with lines lw 2 lt rgb "blue"