def MakeFigures():
"""Make scatterplots.
"""
thinkstats2.RandomSeed(17)
df = brfss.ReadBrfss(nrows=None)
sample = SampleRows(df, 5000, replace=False)
heights, weights = GetHeightWeight(sample)
assert (heights.values[100] == 175)
assert (weights.values[100] == 86.36)
ScatterPlot('brfss_scatter1', heights, weights)
heights, weights = GetHeightWeight(sample, hjitter=1.5, wjitter=1.1)
assert (int(heights.values[100]) == 173)
assert (int(weights.values[100]) == 85)
ScatterPlot('brfss_scatter2', heights, weights)
ScatterPlot('brfss_scatter3', heights, weights, alpha=0.1)
# make a hexbin of all records
heights, weights = GetHeightWeight(df, hjitter=1.3, wjitter=1.1)
assert (int(heights.values[100]) == 171)
assert (int(weights.values[100]) == 55)
HexBin('brfss_scatter4', heights, weights)
def main(script):
thinkstats2.RandomSeed(17)
df = brfss.ReadBrfss(nrows=None)
df = df.dropna(subset=['htm3', 'wtkg2'])
Correlations(df)
return
MakeFigures(df)
BinnedPercentiles(df)
def main(script):
random.seed(100)
np.random.seed(100)
df = brfss.ReadBrfss(nrows=None)
df = df.dropna(subset=["htm3", "wtkg2"])
Correlations(df)
MakeFigures(df)
BinnedPercentiles(df)
def ReadHeights(nrows=None):
"""Read the BRFSS dataset, extract the heights and pickle them.
nrows: number of rows to read
"""
resp = brfss.ReadBrfss(nrows=nrows).dropna(subset=['sex', 'htm3'])
groups = resp.groupby('sex')
d = {}
for name, group in groups:
d[name] = group.htm3.values
return d
def main(name, nrows=None):
thinkstats2.RandomSeed(17)
if nrows is not None:
nrows = int(nrows)
df = brfss.ReadBrfss(nrows=nrows)
columns = df[['htm3', 'wtkg2']].dropna()
heights, weights = columns.htm3.values, columns.wtkg2.values
TestCorrelation(heights, weights)
if nrows == None:
ComputeCorrelations(heights, weights)
def ComputeSkewnesses():
"""Plots KDE of birthweight and adult weight.
"""
def VertLine(x, y):
thinkplot.Plot([x, x], [0, y], color='0.6', linewidth=1)
live, firsts, others = first.MakeFrames()
data = live.totalwgt_lb.dropna()
print('Birth weight')
mean, median = Summarize(data)
y = 0.35
VertLine(mean, y)
thinkplot.Text(mean - 0.15, 0.1 * y, 'mean', horizontalalignment='right')
VertLine(median, y)
thinkplot.Text(median + 0.1, 0.1 * y, 'median', horizontalalignment='left')
pdf = thinkstats2.EstimatedPdf(data)
thinkplot.Pdf(pdf, label='birth weight')
thinkplot.Save(root='density_totalwgt_kde', xlabel='lbs', ylabel='PDF')
df = brfss.ReadBrfss(nrows=None)
data = df.wtkg2.dropna()
print('Adult weight')
mean, median = Summarize(data)
y = 0.02499
VertLine(mean, y)
thinkplot.Text(mean + 1, 0.1 * y, 'mean', horizontalalignment='left')
VertLine(median, y)
thinkplot.Text(median - 1.5,
0.1 * y,
'median',
horizontalalignment='right')
pdf = thinkstats2.EstimatedPdf(data)
thinkplot.Pdf(pdf, label='adult weight')
thinkplot.Save(root='density_wtkg2_kde',
xlabel='kg',
ylabel='PDF',
xlim=[0, 200])
def ComputeSkewnesses():
def VertLine(x, y):
myplots.Plot([x, x], [0, y], color="0.6", linewidth=1)
live, firsts, others = first.MakeFrames()
data = live.totalwgt_lb.dropna()
print("Birth weight")
mean, median = Summarize(data)
y = 0.35
VertLine(mean, y)
myplots.Text(mean - 0.15, 0.1 * y, "mean", horizontalalignment="right")
VertLine(median, y)
myplots.Text(median + 0.1, 0.1 * y, "median", horizontalalignment="left")
pdf = mystats.EstimatedPdf(data)
myplots.Pdf(pdf, label="birth weight")
myplots.Save(root="density_totalwgt_kde", xlabel="lbs", ylabel="PDF")
df = brfss.ReadBrfss(nrows=None)
data = df.wtkg2.dropna()
print("Adult weight")
mean, median = Summarize(data)
y = 0.02499
VertLine(mean, y)
myplots.Text(mean + 1, 0.1 * y, "mean", horizontalalignment="left")
VertLine(median, y)
myplots.Text(median - 1.5, 0.1 * y, "median", horizontalalignment="right")
pdf = mystats.EstimatedPdf(data)
myplots.Pdf(pdf, label="adult weight")
myplots.Save(root="density_wtkg2_kde",
xlabel="kg",
ylabel="PDF",
xlim=[0, 200])
thinkplot.Config(xlable='birth weight (pounds)', ylable='PDF')
#%%
mean = RawMoment(data, 1)
print("mean: {:.2f} pounds".format(mean))
medi = Medinan(data)
print("median: {:.2f} pounds".format(medi))
skewness = Skewness(data)
print("skewness: {:.2f} ".format(skewness))
pearson = PearsonMedianSkewness(data)
print("pearson's median skewness: {:.2f}".format(pearson))
#%%
# BRFSS
import brfss
df = brfss.ReadBrfss(nrows=None)
data = df.wtkg2.dropna()
pdf = thinkstats2.EstimatedPdf(data)
thinkplot.Pdf(pdf, label="adult weight")
thinkplot.Config(xlable='weight (kg)', ylable='PDF')
#%%
pdf = thinkstats2.EstimatedPdf(data)
thinkplot.Pdf(pdf, label="adult weight")
thinkplot.Config(xlable='weight (kg)', ylable='PDF', xlim=[0, 200])
#%%
mean = RawMoment(data, 1)
print("mean: {:.1f} kg".format(mean))
medi = Medinan(data)
print("median: {:.1f} kg".format(medi))
thinkplot.Plot(xs, ys, label='all live')
xs, ys = thinkstats2.NormalProbability(term_weights)
thinkplot.Plot(xs, ys, label='full term')
thinkplot.Config(title='Normal probability plot',
xlabel='Standard deviations from mean',
ylabel='Birth weight (lbs)')
#%% [markdown]
# ## Lognormal model
#
# As an example of a lognormal disrtribution, we'll look at adult weights from the BRFSS.
#%%
import brfss
df = brfss.ReadBrfss()
weights = df.wtkg2.dropna()
#%% [markdown]
# The following function estimates the parameters of a normal distribution and plots the data and a normal model.
#%%
def MakeNormalModel(weights):
"""Plots a CDF with a Normal model.
weights: sequence
"""
cdf = thinkstats2.Cdf(weights, label='weights')
mean, var = thinkstats2.TrimmedMeanVar(weights)
def main(script):
random.seed(100)
np.random.seed(100)
df = brfss.ReadBrfss(nrows=None)
df = df.dropna(subset=["htm3", "wtkg2"])
class DiffMeansResample(h0.DiffMeansPermute):
def RunModel(self):
"""
Goal: Use resampling to simulate test data
Output: Simulated data
"""
group1 = np.random.choice(self.pool, self.n, replace=True)
group2 = np.random.choice(self.pool, self.m, replace=True)
data = group1, group2
return data
preg = nsfg.ReadFemPreg()
live = preg[preg["outcome"] == 1]
resp = nsfg.ReadFemResp()
bs = brfss.ReadBrfss()
income = hinc.ReadData()
log_intp_income = hinc2.InterpolateSample(income, log_upper=6.0)
# Q1. Think Stats Chapter 2 Exercise 4 (effect size of Cohen's d)
first_wt = live.loc[preg["birthord"] == 1, "totalwgt_lb"]
other_wt = live.loc[preg["birthord"] != 1, "totalwgt_lb"]
ts.CohenEffectSize(first_wt, other_wt)
# Q2. Think Stats Chapter 3 Exercise 1 (actual vs. biased)
d = np.diff(np.unique(resp["numkdhh"])).min()
left_of_first_bin = resp["numkdhh"].min() - float(d) / 2
right_of_last_bin = resp["numkdhh"].max() + float(d) / 2
plt.clf()
plt.hist(resp["numkdhh"],
bins=np.arange(left_of_first_bin, right_of_last_bin + d, d),
