plot_arr = plot_arr[np.where(plot_arr[:, 2] > z_thresh)]
print(plot_arr.shape)
x_range = range(0, 101)
x_dict = dict()
for x_val in x_range:
x_dict[x_val] = list()
list_vals = plot_arr[np.where((plot_arr[:, 0]*100.0).astype(np.int16) == x_val)].tolist()
if len(list_vals) > bin_limit:
temp_list = list(range(len(list_vals)))
list_locs = Sublist(temp_list).random_selection(bin_limit)
else:
list_locs = list(range(len(list_vals)))
x_dict[x_val] += list(list_vals[ii] for ii in list_locs)
x_counts = dict()
plot_dicts_ = list()
for k, v in x_dict.items():
x_counts[k] = len(v)
plot_dicts_ += list({xvar: elem[0], yvar: elem[1], zvar: elem[2]} for elem in v)
print(x_counts)
print(len(plot_dicts_))
freq_list, bin_edges = np.histogram(list(float(pt[xvar]) for pt in plot_dicts_), bins=eq_bins, range=(0, 1))
def quantile_regress(x,
y,
q=None,
degree=1,
xlim=None,
ylim=None,
print_summary=False):
if q is None:
q = [0.1, 0.5, 0.9]
elif type(q).__name__ in ('int', 'float', 'str'):
try:
q = [float(q)]
except Exception as e:
print(e)
elif type(q).__name__ in ('list', 'tuple'):
q = Sublist(q)
else:
raise ValueError("Data type not understood: q")
if type(x).__name__ in ('list', 'tuple', 'NoneType'):
x_ = np.array(x)
else:
x_ = x.copy()
if type(y).__name__ in ('list', 'tuple', 'NoneType'):
y_ = np.array(y)
else:
y_ = y.copy()
if xlim is not None:
y_ = y_[np.where((x_ >= xlim[0]) & (x_ <= xlim[1]))]
x_ = x_[np.where((x_ >= xlim[0]) & (x_ <= xlim[1]))]
if ylim is not None:
x_ = x_[np.where((y_ >= ylim[0]) & (y_ <= ylim[1]))]
y_ = y_[np.where((y_ >= ylim[0]) & (y_ <= ylim[1]))]
'''
y_ = y_[:, np.newaxis]
const_ = (x_ * 0.0) + 1.0
x_ = np.hstack([x_[:, np.newaxis], const_[:, np.newaxis]])
'''
df = pd.DataFrame({'y': y_, 'x': x_})
if degree == 0:
raise RuntimeError('Degree of polynomial should be > 0')
rel_str = 'y ~ '
for ii in range(degree):
if ii == 0:
rel_str += 'x'
else:
rel_str += ' + I(x ** {})'.format(str(float(ii + 1)))
mod = smf.quantreg(rel_str, df)
mod.initialize()
q_res = list()
for q_ in q:
res_ = mod.fit(q=q_)
if print_summary:
print(res_.summary())
print('\n')
q_res.append({
'q': q_,
'rsq': res_.rsquared,
'adj_rsq': res_.rsquared_adj,
'coeffs': list(reversed(list(res_.params))),
'pvals': res_.pvalues.tolist(),
'stderrs': res_.bse.tolist()
})
return q_res
plot_arr = plot_arr[np.where(plot_arr[:, 2] > z_thresh)]
print(plot_arr.shape)
x_range = range(0, 101)
x_dict = dict()
for x_val in x_range:
x_dict[x_val] = list()
list_vals = plot_arr[np.where(plot_arr[:, 0] == float(x_val) /
100.0)].tolist()
if len(list_vals) > bin_limit:
temp_list = list(range(len(list_vals)))
list_locs = Sublist(temp_list).random_selection(bin_limit)
else:
list_locs = list(range(len(list_vals)))
x_dict[x_val] += list(list_vals[ii] for ii in list_locs)
x_counts = dict()
plot_dicts_ = list()
for k, v in x_dict.items():
x_counts[k] = len(v)
plot_dicts_ += list({xvar: elem[0], yvar: elem[1]} for elem in v)
print(x_counts)
print(len(plot_dicts_))
freq_list, bin_edges = np.histogram(list(
fire_types = ['single', 'multiple']
xvar = 'tree_cover'
yvar = 'time_since_fire'
zvar = 'decid_frac'
# ylim = (0, 80) # time since fire
ylim = (1950, 2018)
xlim = (0, 1) # tree cover
zlim = (0, 1) # deciduous fraction
zcount_lim = 100000
x_ = np.array(list(range(0, 101))) / 100.0
# y_ = np.array(Sublist.frange(0, 70, 1))
y_ = np.array(Sublist.frange(1950, 2018, 1))
lenx = x_.shape[0]
leny = y_.shape[0]
print(x_)
print(y_)
years = [1992, 2000, 2005, 2010, 2015]
version = 11
filelist = list(in_dir + 'year_{}_{}_{}_fire.csv'.format(
str(year_edge[0])[2:],
str(year_edge[1])[2:], fire_type) for year_edge in year_edges
for fire_type in fire_types)
xvar = '{}_2'.format(bandname)
xlabel = 'Season 2 {}'.format(bandname.upper())
yvar = '{}_1'.format(bandname)
ylabel = 'Season 1 {}'.format(bandname.upper())
pts = list()
for i in range(0, len(val_dicts)):
try:
pts.append((float(val_dicts[i][xvar]), float(val_dicts[i][yvar])))
except ValueError:
pass
maxx = Sublist.percentile(list(elem[0] for elem in pts), 99.995)
maxy = Sublist.percentile(list(elem[1] for elem in pts), 99.995)
max_elem = max(maxx, maxy)
Opt.cprint('X and Y limits: {:3.3f} {:3.3f}'.format(maxx, maxy))
plot_heatmap = {
'type': 'rheatmap2',
'points': pts,
'xlabel': xlabel,
'ylabel': ylabel,
'color_bar_label': 'Data-points per bin',
'plotfile': plot_file,
'xlim': (0, max_elem),
'ylim': (0, max_elem),
fire_types = ['single', 'multiple']
xvar = 'tree_cover'
yvar = 'time_since_fire'
zvar = 'decid_frac'
ylim = (0, 80) # time since fire
# ylim = (1950, 2018)
xlim = (0, 1) # tree cover
zlim = (0, 1) # deciduous fraction
zcount_lim = 100000
x_ = np.array(list(range(0, 101))) / 100.0
# y_ = np.array(Sublist.frange(0, 70, 1))
y_ = np.array(Sublist.frange(0, 80, 1))
lenx = x_.shape[0]
leny = y_.shape[0]
print(x_)
print(y_)
years = [1992, 2000, 2005, 2010, 2015]
version = 12
filelist = list(in_dir + 'year_{}_{}_{}_fire.csv'.format(
str(year_edge[0])[2:],
str(year_edge[1])[2:], fire_type) for year_edge in year_edges
for fire_type in fire_types)
for x_val in x_range:
x_dict[x_val] = list()
list_vals = list()
for dict_ in val_dicts:
if dict_[xvar] == x_val and dict_[yvar] > 0 and dict_[
xvar_] >= cutoff:
temp = {
xvar: float(dict_[xvar]) / 100.0,
yvar: float(dict_[yvar]) / 1000.0,
xvar_: float(dict_[xvar_] / 100.0)
}
list_vals.append(temp)
if len(list_vals) > bin_limit:
list_vals = Sublist(list_vals).random_selection(bin_limit)
x_dict[x_val] += list_vals
x_counts = dict()
plot_dicts_ = list()
for k, v in x_dict.items():
x_counts[k] = len(v)
plot_dicts_ += v
print(x_counts)
print(len(plot_dicts_))
freq_list, bin_edges = np.histogram(list(
float(pt[xvar]) for pt in plot_dicts_),
bins=25,
