ASCII Plots Tutorial¶

This tutorial demonstrates the text-based plotting functions in balance. ASCII plots are useful in terminals, CI logs, notebooks with limited rendering, or anywhere you want a quick visual comparison without a graphical backend.

We cover:

  1. Grouped barplots (ascii_plot_bar) for categorical variables
  2. Grouped histograms (ascii_plot_hist) for numeric variables
  3. Comparative histograms (ascii_comparative_hist) with baseline-relative rendering
  4. ascii_plot_dist — the all-in-one dispatcher
  5. Using library="balance" with the .covars().plot() API
  6. Switching to grouped-bar histograms with comparative=False

Setup¶

Load the toy dataset and run an IPW adjustment so we have three datasets to compare: unadjusted sample, adjusted sample, and target population.

In [1]:
import warnings
warnings.filterwarnings("ignore")

from balance import Sample, load_data

target_df, sample_df = load_data()

sample = Sample.from_frame(sample_df, outcome_columns=["happiness"])
target = Sample.from_frame(target_df, outcome_columns=["happiness"])
sample_with_target = sample.set_target(target)
adjusted = sample_with_target.adjust()

INFO (2026-04-19 15:14:27,504) [__init__/<module> (line 75)]: Using balance version 0.19.0

INFO (2026-04-19 15:14:27,505) [__init__/<module> (line 80)]: 
balance (Version 0.19.0) loaded:
    📖 Documentation: https://import-balance.org/
    🛠️ Help / Issues: https://github.com/facebookresearch/balance/issues/
    📄 Citation:
        Sarig, T., Galili, T., & Eilat, R. (2023).
        balance - a Python package for balancing biased data samples.
        https://arxiv.org/abs/2307.06024

    Tip: You can view this message anytime with balance.help()


WARNING (2026-04-19 15:14:27,527) [input_validation/guess_id_column (line 336)]: Guessed id column name id for the data

WARNING (2026-04-19 15:14:27,540) [sample_frame/from_frame (line 327)]: No weights passed. Adding a 'weight' column and setting all values to 1

WARNING (2026-04-19 15:14:27,543) [input_validation/guess_id_column (line 336)]: Guessed id column name id for the data

WARNING (2026-04-19 15:14:27,557) [sample_frame/from_frame (line 327)]: No weights passed. Adding a 'weight' column and setting all values to 1

INFO (2026-04-19 15:14:27,567) [ipw/ipw (line 723)]: Starting ipw function

INFO (2026-04-19 15:14:27,569) [adjustment/apply_transformations (line 433)]: Adding the variables: []

INFO (2026-04-19 15:14:27,569) [adjustment/apply_transformations (line 434)]: Transforming the variables: ['gender', 'age_group', 'income']

INFO (2026-04-19 15:14:27,578) [adjustment/apply_transformations (line 469)]: Final variables in output: ['gender', 'age_group', 'income']

INFO (2026-04-19 15:14:27,703) [ipw/ipw (line 797)]: Building model matrix

INFO (2026-04-19 15:14:27,704) [ipw/ipw (line 798)]: The formula used to build the model matrix: ['income + gender + age_group + _is_na_gender']

INFO (2026-04-19 15:14:27,704) [ipw/ipw (line 799)]: The number of columns in the model matrix: 16

INFO (2026-04-19 15:14:27,705) [ipw/ipw (line 800)]: The number of rows in the model matrix: 11000

INFO (2026-04-19 15:14:46,049) [ipw/ipw (line 996)]: Done with sklearn

INFO (2026-04-19 15:14:46,050) [ipw/ipw (line 998)]: max_de: None

INFO (2026-04-19 15:14:46,051) [ipw/ipw (line 1020)]: Starting model selection

INFO (2026-04-19 15:14:46,054) [ipw/ipw (line 1075)]: Chosen lambda: 0.041158338186664825

INFO (2026-04-19 15:14:46,055) [ipw/ipw (line 1092)]: Proportion null deviance explained 0.172637976731583

1. All covariates at a glance¶

The easiest way to get an ASCII comparison is through the standard .covars().plot() API with library="balance". This automatically classifies each variable as categorical or numeric and renders grouped barplots or histograms accordingly.

The three fill characters distinguish the datasets:

  • █ = sample (unadjusted)
  • ▒ = adjusted (after IPW)
  • ▐ = population (target)
In [2]:
adjusted.covars().plot(library="balance");

=== gender (categorical) ===

Category | population  adjusted  sample
         |
Female   | ██████████████████████████████████████████ (50.0%)
         | ▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒ (44.5%)
         | ▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐ (29.4%)

Male     | ██████████████████████████████████████████ (50.0%)
         | ▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒ (55.5%)
         | ▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐ (70.6%)

Legend: █ population  ▒ adjusted  ▐ sample
Bar lengths are proportional to weighted frequency within each dataset.

=== age_group (categorical) ===

Category | population  adjusted  sample
         |
18-24    | ████████████████████████ (19.7%)
         | ▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒ (28.1%)
         | ▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐ (49.1%)

25-34    | ████████████████████████████████████ (29.7%)
         | ▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒ (30.7%)
         | ▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐ (30.0%)

35-44    | █████████████████████████████████████ (29.9%)
         | ▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒ (27.4%)
         | ▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐ (15.6%)

45+      | █████████████████████████ (20.6%)
         | ▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒ (13.8%)
         | ▐▐▐▐▐▐ (5.3%)

Legend: █ population  ▒ adjusted  ▐ sample
Bar lengths are proportional to weighted frequency within each dataset.

=== income (numeric, comparative) ===

Range            | population (%) | adjusted (%)   | sample (%)       
----------------------------------------------------------------------
[0.00, 8.57)     | ████████ 49.0  | ████████▒ 54.8 | ████████▒▒▒▒ 73.2
[8.57, 17.14)    | ████ 23.1      | ████ 26.3      | ███] 19.2        
[17.14, 25.71)   | ██ 13.2        | ██ 12.3        | █] 5.3           
[25.71, 34.28)   | █ 7.3          | █ 3.9          | ] 1.6            
[34.28, 42.85)   | █ 3.9          | ] 1.5          | ] 0.4            
[42.85, 51.41)   | 1.8            | 0.2            | 0.1              
[51.41, 59.98)   | 0.9            | 1.0            | 0.2              
[59.98, 68.55)   | 0.4            | 0.0            | 0.0              
[68.55, 77.12)   | 0.2            | 0.0            | 0.0              
[77.12, 85.69)   | 0.1            | 0.0            | 0.0              
[85.69, 94.26)   | 0.0            | 0.0            | 0.0              
[94.26, 102.83)  | 0.0            | 0.0            | 0.0              
[102.83, 111.40) | 0.0            | 0.0            | 0.0              
[111.40, 119.97) | 0.0            | 0.0            | 0.0              
[119.97, 128.54] | 0.0            | 0.0            | 0.0              
----------------------------------------------------------------------
Total            | 100.0          | 100.0          | 100.0            

Key: █ = shared with population, ▒ = excess,    ] = deficit

2. Category spacing¶

By default, barplots insert a blank line between categories for readability. You can disable this with separate_categories=False to get a more compact view.

In [3]:
# Compact view (no blank lines between categories)
adjusted.covars().plot(
    library="balance", variables=["gender"],
    separate_categories=False,
);

=== gender (categorical) ===

Category | population  adjusted  sample
         |
Female   | ██████████████████████████████████████████ (50.0%)
         | ▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒ (44.5%)
         | ▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐ (29.4%)
Male     | ██████████████████████████████████████████ (50.0%)
         | ▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒ (55.5%)
         | ▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐ (70.6%)

Legend: █ population  ▒ adjusted  ▐ sample
Bar lengths are proportional to weighted frequency within each dataset.
In [4]:
# Default view (blank lines between categories)
adjusted.covars().plot(
    library="balance", variables=["gender"],
);

=== gender (categorical) ===

Category | population  adjusted  sample
         |
Female   | ██████████████████████████████████████████ (50.0%)
         | ▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒ (44.5%)
         | ▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐ (29.4%)

Male     | ██████████████████████████████████████████ (50.0%)
         | ▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒ (55.5%)
         | ▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐ (70.6%)

Legend: █ population  ▒ adjusted  ▐ sample
Bar lengths are proportional to weighted frequency within each dataset.

3. Focusing on a single variable¶

Pass variables=["income"] to plot just one covariate. For numeric variables the output is a grouped histogram where each bin shows the weighted proportion for the unadjusted sample, adjusted sample, and target population.

In [5]:
adjusted.covars().plot(
    library="balance", variables=["income"],
);

=== income (numeric, comparative) ===

Range            | population (%) | adjusted (%)   | sample (%)       
----------------------------------------------------------------------
[0.00, 8.57)     | ████████ 49.0  | ████████▒ 54.8 | ████████▒▒▒▒ 73.2
[8.57, 17.14)    | ████ 23.1      | ████ 26.3      | ███] 19.2        
[17.14, 25.71)   | ██ 13.2        | ██ 12.3        | █] 5.3           
[25.71, 34.28)   | █ 7.3          | █ 3.9          | ] 1.6            
[34.28, 42.85)   | █ 3.9          | ] 1.5          | ] 0.4            
[42.85, 51.41)   | 1.8            | 0.2            | 0.1              
[51.41, 59.98)   | 0.9            | 1.0            | 0.2              
[59.98, 68.55)   | 0.4            | 0.0            | 0.0              
[68.55, 77.12)   | 0.2            | 0.0            | 0.0              
[77.12, 85.69)   | 0.1            | 0.0            | 0.0              
[85.69, 94.26)   | 0.0            | 0.0            | 0.0              
[94.26, 102.83)  | 0.0            | 0.0            | 0.0              
[102.83, 111.40) | 0.0            | 0.0            | 0.0              
[111.40, 119.97) | 0.0            | 0.0            | 0.0              
[119.97, 128.54] | 0.0            | 0.0            | 0.0              
----------------------------------------------------------------------
Total            | 100.0          | 100.0          | 100.0            

Key: █ = shared with population, ▒ = excess,    ] = deficit

4. Grouped histogram (ascii_plot_hist)¶

ascii_plot_hist provides a side-by-side view: each dataset gets its own bar (with a distinct fill character) so you can compare how mass is distributed across bins. Bar lengths are proportional to weighted frequency within each dataset.

In [6]:
from balance.stats_and_plots.ascii_plots import ascii_plot_hist

dfs = [
    {"df": target_df, "weight": None},
    {"df": sample_df, "weight": None},
]
print(ascii_plot_hist(
    dfs, names=["Target", "Sample"], column="income",
))

=== income (numeric) ===

Bin              | Target  Sample
                 |
[0.00, 8.57)     | ███████████████████████████████████ (49.0%)
                 | ▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒ (73.2%)
[8.57, 17.14)    | ████████████████ (23.1%)
                 | ▒▒▒▒▒▒▒▒▒▒▒▒▒▒ (19.2%)
[17.14, 25.71)   | █████████ (13.2%)
                 | ▒▒▒▒ (5.3%)
[25.71, 34.28)   | █████ (7.3%)
                 | ▒ (1.6%)
[34.28, 42.85)   | ███ (3.9%)
                 | . (0.4%)
[42.85, 51.41)   | █ (1.8%)
                 | . (0.1%)
[51.41, 59.98)   | █ (0.9%)
                 | . (0.2%)
[59.98, 68.55)   | . (0.4%)
                 |  (0.0%)
[68.55, 77.12)   | . (0.2%)
                 |  (0.0%)
[77.12, 85.69)   | . (0.1%)
                 |  (0.0%)
[85.69, 94.26)   | . (0.0%)
                 |  (0.0%)
[94.26, 102.83)  | . (0.0%)
                 |  (0.0%)
[102.83, 111.40) |  (0.0%)
                 |  (0.0%)
[111.40, 119.97) |  (0.0%)
                 |  (0.0%)
[119.97, 128.54] | . (0.0%)
                 |  (0.0%)

Legend: █ Target  ▒ Sample
Bar lengths are proportional to weighted frequency within each dataset.

5. Comparative histogram (ascii_comparative_hist)¶

When you want to see how a dataset differs from a baseline, ascii_comparative_hist renders each bin relative to the first dataset:

  • █ (solid) = the proportion shared with the baseline
  • ▒ (shaded) = excess beyond the baseline
  • ] (right bracket) = deficit relative to the baseline
In [7]:
from balance.stats_and_plots.ascii_plots import ascii_comparative_hist

# Target (baseline) vs unadjusted sample
dfs = [
    {"df": target_df, "weight": None},
    {"df": sample_df, "weight": None},
]
print(ascii_comparative_hist(
    dfs, names=["Target", "Sample"], column="income",
))

=== income (numeric, comparative) ===

Range            | Target (%)           | Sample (%)                 
---------------------------------------------------------------------
[0.00, 8.57)     | ███████████████ 49.0 | ███████████████▒▒▒▒▒▒▒ 73.2
[8.57, 17.14)    | ███████ 23.1         | ██████] 19.2               
[17.14, 25.71)   | ████ 13.2            | ██ ] 5.3                   
[25.71, 34.28)   | ██ 7.3               |  ] 1.6                     
[34.28, 42.85)   | █ 3.9                | ] 0.4                      
[42.85, 51.41)   | █ 1.8                | ] 0.1                      
[51.41, 59.98)   | 0.9                  | 0.2                        
[59.98, 68.55)   | 0.4                  | 0.0                        
[68.55, 77.12)   | 0.2                  | 0.0                        
[77.12, 85.69)   | 0.1                  | 0.0                        
[85.69, 94.26)   | 0.0                  | 0.0                        
[94.26, 102.83)  | 0.0                  | 0.0                        
[102.83, 111.40) | 0.0                  | 0.0                        
[111.40, 119.97) | 0.0                  | 0.0                        
[119.97, 128.54] | 0.0                  | 0.0                        
---------------------------------------------------------------------
Total            | 100.0                | 100.0                      

Key: █ = shared with Target, ▒ = excess,    ] = deficit

Three-way comparative histogram¶

You can also compare the target, unadjusted, and adjusted distributions side by side. Here the target is the baseline, so ▒ shows where the other datasets have more mass and ] shows where they have less.

In [8]:
dfs = [
    {"df": target.covars().df, "weight": target.weight_series},
    {"df": sample_with_target.covars().df, "weight": sample_with_target.weight_series},
    {"df": adjusted.covars().df, "weight": adjusted.weight_series},
]
print(ascii_comparative_hist(
    dfs, names=["Target", "Unadjusted", "Adjusted"],
    column="income",
))

=== income (numeric, comparative) ===

Range            | Target (%)    | Unadjusted (%)    | Adjusted (%)  
---------------------------------------------------------------------
[0.00, 8.57)     | ████████ 49.0 | ████████▒▒▒▒ 73.2 | ████████▒ 54.8
[8.57, 17.14)    | ████ 23.1     | ███] 19.2         | ████ 26.3     
[17.14, 25.71)   | ██ 13.2       | █] 5.3            | ██ 12.3       
[25.71, 34.28)   | █ 7.3         | ] 1.6             | █ 3.9         
[34.28, 42.85)   | █ 3.9         | ] 0.4             | ] 1.5         
[42.85, 51.41)   | 1.8           | 0.1               | 0.2           
[51.41, 59.98)   | 0.9           | 0.2               | 1.0           
[59.98, 68.55)   | 0.4           | 0.0               | 0.0           
[68.55, 77.12)   | 0.2           | 0.0               | 0.0           
[77.12, 85.69)   | 0.1           | 0.0               | 0.0           
[85.69, 94.26)   | 0.0           | 0.0               | 0.0           
[94.26, 102.83)  | 0.0           | 0.0               | 0.0           
[102.83, 111.40) | 0.0           | 0.0               | 0.0           
[111.40, 119.97) | 0.0           | 0.0               | 0.0           
[119.97, 128.54] | 0.0           | 0.0               | 0.0           
---------------------------------------------------------------------
Total            | 100.0         | 100.0             | 100.0         

Key: █ = shared with Target, ▒ = excess,    ] = deficit

6. Grouped histograms via comparative=False¶

By default, ascii_plot_dist (and .covars().plot(library="balance")) renders numeric variables as comparative histograms showing excess/deficit relative to a baseline. If you prefer the simpler grouped-bar style (same layout used for categorical variables), pass comparative=False.

In [9]:
# Comparative mode (default) — numeric variables show excess/deficit vs baseline
adjusted.covars().plot(
    library="balance", variables=["income"],
);

=== income (numeric, comparative) ===

Range            | population (%) | adjusted (%)   | sample (%)       
----------------------------------------------------------------------
[0.00, 8.57)     | ████████ 49.0  | ████████▒ 54.8 | ████████▒▒▒▒ 73.2
[8.57, 17.14)    | ████ 23.1      | ████ 26.3      | ███] 19.2        
[17.14, 25.71)   | ██ 13.2        | ██ 12.3        | █] 5.3           
[25.71, 34.28)   | █ 7.3          | █ 3.9          | ] 1.6            
[34.28, 42.85)   | █ 3.9          | ] 1.5          | ] 0.4            
[42.85, 51.41)   | 1.8            | 0.2            | 0.1              
[51.41, 59.98)   | 0.9            | 1.0            | 0.2              
[59.98, 68.55)   | 0.4            | 0.0            | 0.0              
[68.55, 77.12)   | 0.2            | 0.0            | 0.0              
[77.12, 85.69)   | 0.1            | 0.0            | 0.0              
[85.69, 94.26)   | 0.0            | 0.0            | 0.0              
[94.26, 102.83)  | 0.0            | 0.0            | 0.0              
[102.83, 111.40) | 0.0            | 0.0            | 0.0              
[111.40, 119.97) | 0.0            | 0.0            | 0.0              
[119.97, 128.54] | 0.0            | 0.0            | 0.0              
----------------------------------------------------------------------
Total            | 100.0          | 100.0          | 100.0            

Key: █ = shared with population, ▒ = excess,    ] = deficit
In [10]:
# Grouped-bar mode — numeric variables use the same bar style as categorical
adjusted.covars().plot(
    library="balance", variables=["income"],
    comparative=False,
);

=== income (numeric) ===

Bin              | population  adjusted  sample
                 |
[0.00, 8.57)     | ███████████████████████████████████ (49.0%)
                 | ▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒ (54.8%)
                 | ▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐▐ (73.2%)
[8.57, 17.14)    | ████████████████ (23.1%)
                 | ▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒ (26.3%)
                 | ▐▐▐▐▐▐▐▐▐▐▐▐▐▐ (19.2%)
[17.14, 25.71)   | █████████ (13.2%)
                 | ▒▒▒▒▒▒▒▒▒ (12.3%)
                 | ▐▐▐▐ (5.3%)
[25.71, 34.28)   | █████ (7.3%)
                 | ▒▒▒ (3.9%)
                 | ▐ (1.6%)
[34.28, 42.85)   | ███ (3.9%)
                 | ▒ (1.5%)
                 | . (0.4%)
[42.85, 51.41)   | █ (1.8%)
                 | . (0.2%)
                 | . (0.1%)
[51.41, 59.98)   | █ (0.9%)
                 | ▒ (1.0%)
                 | . (0.2%)
[59.98, 68.55)   | . (0.4%)
                 |  (0.0%)
                 |  (0.0%)
[68.55, 77.12)   | . (0.2%)
                 |  (0.0%)
                 |  (0.0%)
[77.12, 85.69)   | . (0.1%)
                 |  (0.0%)
                 |  (0.0%)
[85.69, 94.26)   | . (0.0%)
                 |  (0.0%)
                 |  (0.0%)
[94.26, 102.83)  | . (0.0%)
                 |  (0.0%)
                 |  (0.0%)
[102.83, 111.40) |  (0.0%)
                 |  (0.0%)
                 |  (0.0%)
[111.40, 119.97) |  (0.0%)
                 |  (0.0%)
                 |  (0.0%)
[119.97, 128.54] | . (0.0%)
                 |  (0.0%)
                 |  (0.0%)

Legend: █ population  ▒ adjusted  ▐ sample
Bar lengths are proportional to weighted frequency within each dataset.