def plot_param_matrix(
plotter: StreamlitPlotter,
calib_dir_path: str,
mcmc_tables: List[pd.DataFrame],
mcmc_params: List[pd.DataFrame],
targets: dict,
app_name: str,
region: str,
show_ticks=False,
):
parameters = mcmc_params[0]["name"].unique().tolist()
chain_length = find_shortest_chain_length(mcmc_tables)
burn_in = st.sidebar.slider("Burn in", 0, chain_length, 0)
label_font_size = st.sidebar.slider("Label font size", 1, 15, 8)
label_chars = st.sidebar.slider("Label characters", 1, 10, 2)
bins = st.sidebar.slider("Bins", 4, 50, 20)
style = st.sidebar.selectbox("Style", ["Shade", "Scatter", "KDE"])
dpi_request = st.sidebar.slider("DPI", 50, 2000, 300)
plots.calibration.plots.plot_param_vs_param(
plotter,
mcmc_params,
parameters,
burn_in,
style,
bins,
label_font_size,
label_chars,
dpi_request,
show_ticks=show_ticks,
)
st.write(parameters)
def plot_posterior(
plotter: StreamlitPlotter,
calib_dir_path: str,
mcmc_tables: List[pd.DataFrame],
mcmc_params: List[pd.DataFrame],
targets: dict,
app_name: str,
region: str,
):
chosen_param = selectors.parameter(mcmc_params[0])
chain_length = find_shortest_chain_length(mcmc_tables)
burn_in = st.sidebar.slider("Burn in", 0, chain_length, 0)
num_bins = st.sidebar.slider("Number of bins", 1, 50, 16)
prior = None
priors = []
try:
priors_path = os.path.join(calib_dir_path, "priors-1.yml")
with open(priors_path) as file:
priors = yaml.load(file, Loader=yaml.FullLoader)
except:
st.write("Check if priors-1.yml exists in the output folder")
for i, prior in enumerate(priors):
if prior["param_name"] == chosen_param:
prior = priors[i]
break
plots.calibration.plots.plot_posterior(plotter, mcmc_params, mcmc_tables,
burn_in, chosen_param, num_bins,
prior)
def plot_loglike_vs_all_params(
plotter: StreamlitPlotter,
calib_dir_path: str,
mcmc_tables: List[pd.DataFrame],
mcmc_params: List[pd.DataFrame],
targets: dict,
app_name: str,
region: str,
):
(
title_font_size,
label_font_size,
dpi_request,
capitalise_first_letter,
) = selectors.create_standard_plotting_sidebar()
chain_length = find_shortest_chain_length(mcmc_tables)
burn_in = st.sidebar.slider("Burn in", 0, chain_length, 0)
plots.calibration.plots.plot_all_params_vs_loglike(
plotter,
mcmc_tables,
mcmc_params,
burn_in,
title_font_size,
label_font_size,
capitalise_first_letter,
dpi_request,
)
def plot_calibration_fit(
plotter: StreamlitPlotter,
calib_dir_path: str,
mcmc_tables: List[pd.DataFrame],
mcmc_params: List[pd.DataFrame],
targets: dict,
app_name: str,
region: str,
):
# Set up interface
available_outputs = [o["output_key"] for o in targets.values()]
chain_length = find_shortest_chain_length(mcmc_tables)
burn_in = st.sidebar.slider(
"Burn in (select 0 for default behaviour of discarding first half)", 0,
chain_length, 0)
chosen_output = st.sidebar.selectbox("Select calibration target",
available_outputs)
is_logscale = st.sidebar.checkbox("Log scale")
# Get data for plotting
outputs = get_uncertainty_data(calib_dir_path, mcmc_tables, chosen_output,
burn_in)
# Call main plotting function
plots.calibration.plots.plot_calibration_fit(plotter, chosen_output,
outputs, targets, is_logscale)
def plot_mcmc_parameter_trace(
plotter: StreamlitPlotter,
calib_dir_path: str,
mcmc_tables: List[pd.DataFrame],
mcmc_params: List[pd.DataFrame],
targets: dict,
app_name: str,
region: str,
):
chosen_param = selectors.parameter(mcmc_params[0])
chain_length = find_shortest_chain_length(mcmc_tables)
burn_in = st.sidebar.slider("Burn in", 0, chain_length, 0)
plots.calibration.plots.plot_mcmc_parameter_trace(plotter, mcmc_params,
burn_in, chosen_param)
def plot_all_posteriors(
plotter: StreamlitPlotter,
calib_dir_path: str,
mcmc_tables: List[pd.DataFrame],
mcmc_params: List[pd.DataFrame],
targets: dict,
app_name: str,
region: str,
):
priors = []
try:
priors_path = os.path.join(calib_dir_path, "priors-1.yml")
with open(priors_path) as file:
priors = yaml.load(file, Loader=yaml.FullLoader)
except:
st.write("Check if priors-1.yml exists in the output folder")
(
title_font_size,
label_font_size,
dpi_request,
capitalise_first_letter,
) = selectors.create_standard_plotting_sidebar()
chain_length = find_shortest_chain_length(mcmc_tables)
burn_in = st.sidebar.slider("Burn in", 0, chain_length, 0)
num_bins = st.sidebar.slider("Number of bins", 1, 50, 16)
sig_figs = st.sidebar.slider("Significant figures", 0, 6, 3)
plots.calibration.plots.plot_multiple_posteriors(
plotter,
mcmc_params,
mcmc_tables,
burn_in,
num_bins,
title_font_size,
label_font_size,
capitalise_first_letter,
dpi_request,
priors,
parameters=get_epi_params(mcmc_params),
)
write_mcmc_centiles(mcmc_params, mcmc_tables, burn_in, sig_figs,
[2.5, 50, 97.5])
def plot_multi_output_fit(
plotter: StreamlitPlotter,
calib_dir_path: str,
mcmc_tables: List[pd.DataFrame],
mcmc_params: List[pd.DataFrame],
targets: dict,
app_name: str,
region: str,
):
# Set up interface
available_outputs = [o["output_key"] for o in targets.values()]
(
title_font_size,
label_font_size,
dpi_request,
capitalise_first_letter,
) = selectors.create_standard_plotting_sidebar()
is_logscale = st.sidebar.checkbox("Log scale")
chain_length = find_shortest_chain_length(mcmc_tables)
burn_in = st.sidebar.slider(
"Burn in (select 0 for default behaviour of discarding first half)", 0,
chain_length, 0)
# Get data for plotting
outputs = {
output: get_uncertainty_data(calib_dir_path, mcmc_tables, output,
burn_in)
for output in available_outputs
}
# Call main plotting function
plots.calibration.plots.plot_multi_fit(
plotter,
available_outputs,
outputs,
targets,
is_logscale,
title_font_size,
label_font_size,
dpi_request,
capitalise_first_letter,
)
def plot_acceptance_ratio(
plotter: StreamlitPlotter,
calib_dir_path: str,
mcmc_tables: List[pd.DataFrame],
mcmc_params: List[pd.DataFrame],
targets: dict,
app_name: str,
region: str,
):
label_font_size = st.sidebar.slider("Label font size", 1, 15, 10)
burn_in = st.sidebar.slider("Burn in", 0,
find_shortest_chain_length(mcmc_tables), 0)
dpi_request = st.sidebar.slider("DPI", 50, 2000, 300)
plots.calibration.plots.plot_acceptance_ratio(
plotter,
mcmc_tables,
burn_in,
label_font_size=label_font_size,
dpi_request=dpi_request)