alternative visualisations

biocypher · Aug 22, 2024 · 9296287 · 9296287
1 parent 6bb6072
commit 9296287
Show file tree

Hide file tree

Showing 16 changed files with 69 additions and 18 deletions.
diff --git a/docs/images/boxplot-naive-vs-biochatter.pdf b/docs/images/boxplot-naive-vs-biochatter.pdf
diff --git a/docs/images/dotplot-per-task.pdf b/docs/images/dotplot-per-task.pdf
diff --git a/docs/images/dotplot-per-task.png b/docs/images/dotplot-per-task.png
diff --git a/docs/images/scatter-naive-vs-biochatter.pdf b/docs/images/scatter-naive-vs-biochatter.pdf
diff --git a/docs/images/scatter-naive-vs-biochatter.png b/docs/images/scatter-naive-vs-biochatter.png
diff --git a/docs/images/scatter-per-quantisation-name.pdf b/docs/images/scatter-per-quantisation-name.pdf
diff --git a/docs/images/scatter-per-quantisation-name.png b/docs/images/scatter-per-quantisation-name.png
diff --git a/docs/images/scatter-quantisation-accuracy.pdf b/docs/images/scatter-quantisation-accuracy.pdf
diff --git a/docs/images/scatter-size-accuracy.pdf b/docs/images/scatter-size-accuracy.pdf
diff --git a/docs/images/stripplot-extraction-tasks.png b/docs/images/stripplot-extraction-tasks.png
diff --git a/docs/images/stripplot-per-model.png b/docs/images/stripplot-per-model.png
diff --git a/docs/images/stripplot-rag-tasks.pdf b/docs/images/stripplot-rag-tasks.pdf
diff --git a/docs/images/stripplot-rag-tasks.png b/docs/images/stripplot-rag-tasks.png
diff --git a/docs/images/violin-naive-vs-biochatter.pdf b/docs/images/violin-naive-vs-biochatter.pdf
diff --git a/docs/images/violin-naive-vs-biochatter.png b/docs/images/violin-naive-vs-biochatter.png
diff --git a/docs/scripts/hooks.py b/docs/scripts/hooks.py
@@ -117,9 +117,7 @@ def plot_text2cypher():
  else (
  "llama-3"
  if "llama-3" in x
- else "gpt"
- if "gpt" in x
- else "other open source"
+ else "gpt" if "gpt" in x else "other open source"
  )
  )
  )
@@ -261,9 +259,9 @@ def preprocess_results_for_frontend(
  axis=1,
  )
 
- aggregated_scores[
- "Full model name"
- ] = aggregated_scores.index.get_level_values("model_name")
+ aggregated_scores["Full model name"] = (
+ aggregated_scores.index.get_level_values("model_name")
+ )
  aggregated_scores["Score achieved"] = aggregated_scores["score_achieved"]
  aggregated_scores["Score possible"] = aggregated_scores["score_possible"]
  aggregated_scores["Score SD"] = aggregated_scores["score_sd"]
@@ -333,9 +331,9 @@ def write_individual_extraction_task_results(raw_results: pd.DataFrame) -> None:
  axis=1,
  )
 
- aggregated_scores[
- "Full model name"
- ] = aggregated_scores.index.get_level_values("model_name")
+ aggregated_scores["Full model name"] = (
+ aggregated_scores.index.get_level_values("model_name")
+ )
  aggregated_scores["Subtask"] = aggregated_scores.index.get_level_values(
  "subtask"
  )
@@ -392,9 +390,9 @@ def create_overview_table(result_files_path: str, result_file_names: list[str]):
  )
 
  overview_per_quantisation = overview
- overview_per_quantisation[
- "Full model name"
- ] = overview_per_quantisation.index
+ overview_per_quantisation["Full model name"] = (
+ overview_per_quantisation.index
+ )
  overview_per_quantisation[
  ["Model name", "Size", "Version", "Quantisation"]
  ] = overview_per_quantisation["Full model name"].str.split(":", expand=True)
@@ -428,9 +426,9 @@ def create_overview_table(result_files_path: str, result_file_names: list[str]):
  ]
  ]
  # round mean and sd to 2 decimal places
- overview_per_quantisation.loc[
- :, "Median Accuracy"
- ] = overview_per_quantisation["Median Accuracy"].round(2)
+ overview_per_quantisation.loc[:, "Median Accuracy"] = (
+ overview_per_quantisation["Median Accuracy"].round(2)
+ )
  overview_per_quantisation.loc[:, "SD"] = overview_per_quantisation[
  "SD"
  ].round(2)
@@ -878,9 +876,9 @@ def plot_extraction_tasks():
  axis=1,
  )
 
- aggregated_scores[
- "Full model name"
- ] = aggregated_scores.index.get_level_values("model_name")
+ aggregated_scores["Full model name"] = (
+ aggregated_scores.index.get_level_values("model_name")
+ )
  aggregated_scores["Subtask"] = aggregated_scores.index.get_level_values(
  "subtask"
  )
@@ -1093,6 +1091,9 @@ def plot_comparison_naive_biochatter(overview):
  )
  ]
 
+ # print number of rows of each task
+ print(overview_melted["Task"].value_counts())
+
  sns.set_theme(style="whitegrid")
  plt.figure(figsize=(6, 4))
  sns.boxplot(
@@ -1118,6 +1119,56 @@ def plot_comparison_naive_biochatter(overview):
  )
  plt.close()
 
+ # plot scatter plot
+ plt.figure(figsize=(6, 4))
+ sns.stripplot(
+ x="Task",
+ y="Accuracy",
+ data=overview_melted,
+ jitter=0.2,
+ alpha=0.8,
+ )
+ plt.ylim(0, 1)
+ plt.xlabel(None)
+ plt.xticks(
+ ticks=range(len(overview_melted["Task"].unique())),
+ labels=["BioChatter", "Naive LLM (using full YAML schema)"],
+ )
+ plt.savefig(
+ "docs/images/scatter-naive-vs-biochatter.png",
+ bbox_inches="tight",
+ dpi=300,
+ )
+ plt.savefig(
+ "docs/images/scatter-naive-vs-biochatter.pdf",
+ bbox_inches="tight",
+ )
+ plt.close()
+
+ # plit violin plot
+ plt.figure(figsize=(6, 4))
+ sns.violinplot(
+ x="Task",
+ y="Accuracy",
+ data=overview_melted,
+ )
+ plt.ylim(0, 1)
+ plt.xlabel(None)
+ plt.xticks(
+ ticks=range(len(overview_melted["Task"].unique())),
+ labels=["BioChatter", "Naive LLM (using full YAML schema)"],
+ )
+ plt.savefig(
+ "docs/images/violin-naive-vs-biochatter.png",
+ bbox_inches="tight",
+ dpi=300,
+ )
+ plt.savefig(
+ "docs/images/violin-naive-vs-biochatter.pdf",
+ bbox_inches="tight",
+ )
+ plt.close()
+
 
 def calculate_stats(overview):
  overview_melted = melt_and_process(overview)